search

liuheng92 / tensorflow_PSENet

This is a tensorflow re-implementation of PSENet: Shape Robust Text Detection with Progressive Scale Expansion Network.My blog:
https://blog.csdn.net/liuxiaoheng1992/article/details/87646951
MIT License
489 stars 162 forks source link

Training icdar2017, it seems not converging, what's wrong ? #31

Closed XiaoLaoDi closed 5 years ago

XiaoLaoDi commented 5 years ago

Here is the beginning of the training log.

/root/anaconda3/envs/psenet/bin/python3.5 /data3/xxx/tensorflow_PSENet/train.py --gpu_list=0,1 --input_size=512 --batch_size_per_gpu=4 --checkpoint_path=./resnet_v1_50/ --training_data_path=./icdar2017rctw_train_v1.2/train/ DEBUG:matplotlib:$HOME=/root DEBUG:matplotlib:CONFIGDIR=/root/.config/matplotlib DEBUG:matplotlib:matplotlib data path: /root/anaconda3/envs/psenet/lib/python3.5/site-packages/matplotlib/mpl-data DEBUG:matplotlib:loaded rc file /root/anaconda3/envs/psenet/lib/python3.5/site-packages/matplotlib/mpl-data/matplotlibrc DEBUG:matplotlib:matplotlib version 3.0.3 DEBUG:matplotlib:interactive is False DEBUG:matplotlib:platform is linux DEBUG:matplotlib:loaded modules: ['email.parser', 'tensorflow.contrib.integrate.python.ops', '_collections_abc', 'tensorflow.python.ops.random_ops', 'pkg_resources.extern.packaging.specifiers', 'numpy.polynomial.hermite', 'tensorflow.contrib.distributions.python.ops.bijectors.conditional_bijector_impl', 'tensorflow.contrib.learn.python.learn.datasets.synthetic', 'cv2.face', 'google.protobuf.internal._api_implementation', 'unittest.main', 'tensorflow.contrib.labeled_tensor.python.ops.sugar', 'tensorflow.python.saved_model.utils', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.reader_source', 'tensorflow.contrib.opt.python.training.moving_average_optimizer', 'tensorflow.contrib.solvers.python.ops.least_squares', 'cv2.ogl', 'cv2.instr', 'google.protobuf.text_format', 'tensorflow.contrib.keras.api.keras.datasets.mnist', 'tensorflow.contrib.linalg.python.ops.linear_operator_full_matrix', '_sitebuiltins', 'tensorflow.python.ops.confusion_matrix', 'tensorflow.contrib.metrics.python.metrics.classification', 'tensorflow.core.protobuf.saved_model_pb2', 'posix', 'tensorflow.python.layers.layers', 'cv2.motempl', 'encodings.unicode_escape', 'numpy.core._multiarray_umath', 'tensorflow.contrib.distributions.python.ops.onehot_categorical', 'tensorflow.python.estimator.export.export_output', 'tensorflow.contrib.learn.python.learn.preprocessing.text', 'tensorflow.python.training.session_manager', 'tensorflow.contrib.linear_optimizer.python.ops.sparse_feature_column', 'tensorflow.python.saved_model.loader_impl', 'tensorflow.contrib.seq2seq.python.ops.dynamic_attention_wrapper', 'tensorflow.contrib.keras.api.keras.datasets', 'google.protobuf.internal.enum_type_wrapper', '_hashlib', 'tensorflow.contrib.slim', 'tensorflow.contrib.input_pipeline.python.ops.input_pipeline_ops', 'tensorflow.python.framework.cpp_shape_inference_pb2', 'tensorflow.contrib.tensorboard.plugins.projector', 'tensorflow.python.saved_model.builder', 'tensorflow.contrib.session_bundle.gc', 'tensorflow.contrib.cloud.python', 'cv2.img_hash', 'tensorflow.contrib.metrics.python.ops.metric_ops', 'numpy.core.shape_base', 'tensorflow.contrib.rnn.python.ops', 'tensorflow.contrib.training.python.training.device_setter', 'tensorflow.contrib.distributions.python.ops.conditional_distribution', 'tensorflow.contrib.solvers.python.ops.linear_equations', 'tensorflow.contrib.linalg.python.ops.linear_operator_udvh_update', 'tensorflow.python.saved_model.main_op_impl', 'matplotlib.colors', 'tensorflow.python.platform.googletest', 'tensorflow.contrib.training.python.training.training', 'unittest.result', 'tensorflow.contrib.distributions.python.ops.bijectors.affine', 'tensorflow.contrib.stat_summarizer', 'string', 'tensorflow.contrib.keras.python.keras.regularizers', 'tensorflow.contrib.keras.api.keras.initializers', 'tensorflow.python.ops.gen_script_ops', 'tensorflow.python.training.adagrad', 'numpy.lib.histograms', 'tensorflow.contrib.util.loader', 'tensorflow.contrib.learn.python.learn.learn_io.pandas_io', 'tensorflow.contrib.distributions.python.ops.bijectors.invert_impl', 'tensorflow.core.framework.variable_pb2', 'b7e078e27f11ed49d55125fdb2fb1324', '_string', 'unittest.loader', 'cv2.structured_light', 'tensorflow.contrib.cudnn_rnn.ops.gen_cudnn_rnn_ops', 'distutils.version', 'tensorflow.contrib.framework.python.framework.experimental', 'tensorflow.python.training.slot_creator', 'tensorflow.contrib.keras.api.keras.optimizers', 'multiprocessing.context', 'tensorflow.contrib.keras.python.keras.metrics', 'numpy.polynomial', 'numpy.testing', 'tensorflow.python.training.coordinator', 'google.protobuf.internal.containers', 'tensorflow.python.ops.functional_ops', 'tensorflow.contrib.opt.python.training.lazy_adam_optimizer', 'tensorflow.contrib.keras.python.keras.utils.data_utils', 'tensorflow.contrib.distributions.python.ops.mvn_linear_operator', 'atexit', '_sysconfigdata', 'tensorflow.contrib.learn.python.learn.estimators.kmeans', 'tensorflow.contrib.distributions.python.ops.dirichlet', 'tensorflow.contrib.tensor_forest.python.constants', 'tensorflow.contrib.specs', 'tensorflow.python.framework.subscribe', '_lzma', 'numpy.lib._iotools', 'tensorflow.contrib.distributions.python.ops.laplace', 'numpy.config', 'linecache', 'tensorflow.contrib.layers.python.layers.feature_column', 'tensorflow.contrib.distributions.python.ops.shape', 'tensorflow.contrib.tfprof', 'tensorflow.contrib.seq2seq.python.ops.decoder', 'tensorflow.contrib.graph_editor', 'tensorflow.python.ops.math_ops', 'functools', 'inspect', 'tensorflow.python.training.training_util', 'tensorflow.core.framework.op_def_pb2', 'numpy.lib.shape_base', 'textwrap', 'decimal', 'tensorflow.contrib.distributions.python', 'tensorflow.python.framework.framework_lib', 'tensorflow.core.framework.types_pb2', 'tensorflow.contrib.keras.python.keras.utils.vis_utils', 'tensorflow.python.ops.nn_ops', 'nets.resnet.resnet_utils', 'tensorflow.contrib.keras.python.keras.callbacks', 'pkg_resources.extern.six.moves.urllib', 'tensorflow.contrib.opt.python', 'tensorflow.python.layers.normalization', 'tensorflow.contrib.keras.api.keras.preprocessing.image', 'tensorflow.contrib.tensorboard.plugins.projector.projector_config_pb2', 'tensorflow.contrib.linalg.python.ops.linear_operator_identity', 'threading', 'tensorflow.python.summary.writer.writer', 'matplotlib.cbook.deprecation', 'tensorflow.contrib.distributions.python.ops.binomial', 'tensorflow.contrib.keras.api.keras.callbacks', '_json', 'tensorflow.contrib.factorization.python.ops', 'tensorflow.contrib.keras.python.keras.utils.generic_utils', '_functools', 'pprint', 'tensorflow.contrib.distributions.python.ops.bijectors.softmax_centered_impl', 'tensorflow.contrib.factorization.python.ops.gen_factorization_ops', 'tensorflow.contrib.learn.python.learn.estimators', 'site', 'numpy.lib.arrayterator', 'tensorflow.python.platform.flags', 'tensorflow', 'numpy.ma', 'tensorflow.contrib.keras.python.keras.layers.advanced_activations', 'tensorflow.contrib.testing.python', 'tensorflow.contrib.distributions.python.ops.normal', 'tensorflow.contrib.keras.python.keras.applications.xception', 'tensorflow.contrib.linalg.python.ops', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.boolean_mask', 'tensorflow.python.ops.math_grad', 'imp', 'tensorflow.contrib.cudnn_rnn.python', 'tensorflow.contrib.learn.python.learn.ops.losses_ops', 'errno', 'tensorflow.contrib.slim.python', 'email.iterators', '_signal', 'tensorflow.contrib.lookup.lookup_ops', 'mp_main__', 'numpy.core.numerictypes', 'tensorflow.contrib.distributions.python.ops.student_t', 'tensorflow.python.framework.meta_graph', 'tensorflow.contrib.learn.python.learn.utils.export', '78f3f6833449015a20606668f4590bf9', 'tensorflow.contrib.quantization.python.math_ops', 'google.protobuf.pyext.cpp_message', '_bisect', 'tensorflow.contrib.bayesflow.python.ops.monte_carlo', 'tensorflow.contrib.training.python.training.sequence_queueing_state_saver', 'ctypes._endian', 'opcode', 'tensorflow.python.ops.gen_set_ops', 'tensorflow.core.framework', 'numpy.core.info', 'numpy.version', 'tensorflow.contrib.learn.python.learn.estimators.logistic_regressor', 'pkg_resources.extern.packaging._structures', 'numpy.fft.fftpack_lite', 'tensorflow.core.framework.cost_graph_pb2', '_pickle', '_weakrefset', 'tensorflow.contrib.keras.python.keras.layers.convolutional_recurrent', 'genericpath', 'tensorflow.python.training.adagrad_da', 'email.encoders', '_bootlocale', 'email.utils', 'tensorflow.contrib.keras.python.keras.datasets.imdb', 'tensorflow.core.framework.versions_pb2', 'tensorflow.python.training.proximal_gradient_descent', '_io', 'tensorflow.python.framework.versions', 'tensorflow.python.ops.gen_image_ops', 'sre_compile', 'tensorflow.contrib.rnn.python.ops.core_rnn_cell', 'email._parseaddr', 'builtins', 'tensorflow.contrib.sparsemax', 'tensorflow.contrib.session_bundle', 'numpy.polynomial.polyutils', 'tensorflow.python.ops.variable_scope', 'tensorflow.contrib.lookup', 'tensorflow.contrib.keras.python.keras.preprocessing', 'tensorflow.python.ops.metrics', 'tensorflow.contrib.learn.python.learn.preprocessing.categorical', 'nets.model', 'tensorflow.contrib.bayesflow.python.ops.stochastic_gradient_estimators', 'tensorflow.python.ops.gen_spectral_ops', 'tensorflow.core', 'numpy.lib.stride_tricks', 'matplotlib.cbook', 'tensorflow.contrib.solvers.python', 'tensorflow.contrib.legacy_seq2seq', 'tensorflow.contrib.keras.api.keras.regularizers', 'tensorflow.contrib.specs.python.summaries', 'tensorflow.contrib.session_bundle.constants', 'operator', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.batch', 'tensorflow.contrib.opt.python.training.variable_clipping_optimizer', 'pkg_resources.extern.appdirs', 'tensorflow.python.framework.random_seed', 'numpy.core.defchararray', 'tensorflow.python.saved_model.main_op', 'tensorflow.contrib.keras.python.keras.layers.pooling', 'tensorflow.python.training.basic_loops', 'tensorflow.contrib.learn.python.learn.estimators.dnn', 'tensorflow.contrib.keras.python.keras.layers.merge', 'numpy.core.numeric', 'tensorflow.core.protobuf.debug_pb2', 'numpy.core.memmap', 'tensorflow.contrib.keras.api.keras.preprocessing', 'ssl', 'tensorflow.python.ops.ctc_ops', 'sre_constants', 'cv2.rgbd', 'warnings', 'google.protobuf.internal', 'numpy.linalg.lapack_lite', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.sparsify', 'tensorflow.contrib.tensorboard', 'google.protobuf.internal.type_checkers', 'tensorflow.python.saved_model.tag_constants', 'tensorflow.contrib.distributions.python.ops.operator_pd_full', 'tensorflow.contrib.tensor_forest.python.ops', 'email', 'tensorflow.python.training.monitored_session', 'tensorflow.contrib.learn.python.learn.estimators.dynamic_rnn_estimator', 'reprlib', 'tensorflow.contrib.factorization.python.ops.clustering_ops', 'easydict', '_random', 'tensorflow.contrib.keras.api.keras.applications', 'tensorflow.contrib.keras.python.keras.engine.training', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.unary_transforms', 'tensorflow.contrib.seq2seq.python.ops.helper', 'numpy.core._type_aliases', 'numpy.core.records', 'tensorflow.contrib.keras.api.keras.datasets.reuters', 'tensorflow.contrib.testing.python.framework.util_test', 'google.protobuf.reflection', '58660d504b637a90ef22b65c6c82c9ed', 'sre_parse', 'tensorflow.contrib.keras.python.keras.layers.serialization', 'tensorflow.core.protobuf', 'email.header', 'tensorflow.python.ops.gen_ctc_ops', 'tensorflow.core.example.example_pb2', 'tensorflow.contrib.tensor_forest.client.eval_metrics', 'tensorflow.contrib.memory_stats.ops', 'tensorflow.core.protobuf.control_flow_pb2', 'tensorflow.contrib.rnn', '4f6a1f13bcf3a21f78e647afbc3808f7', 'tensorflow.python.framework.op_def_registry', 'tensorflow.contrib.saved_model.python.saved_model', 'multiprocessing', 'numpy.lib.mixins', 'cv2.ximgproc', 'main__', 'tensorflow.core.framework.tensor_shape_pb2', 'tensorflow.core.util', '_datetime', 'tensorflow.contrib.distributions', 'tensorflow.contrib.seq2seq.python', 'tensorflow.contrib.keras.python.keras.backend', 'tensorflow.contrib.keras.python.keras.applications', 'tensorflow.python.client.device_lib', 'unittest.signals', 'six.moves.urllib.request', 'tensorflow.python.platform', 'tensorflow.contrib.cudnn_rnn.python.ops', 'unittest.mock', 'tensorflow.core.lib.core.error_codes_pb2', 'tensorflow.contrib.input_pipeline.python.ops', 'pkgutil', 'tensorflow.contrib.layers.python.ops.sparse_feature_cross_op', 'tensorflow.contrib.rnn.ops', 'tensorflow.python.ops.gen_parsing_ops', 'tensorflow.python.ops.gen_data_flow_ops', 'cv2.xphoto', 'locale', 'tensorflow.contrib.keras.python.keras', 'tensorflow.python.training.summary_io', 'tensorflow.contrib.keras.python.keras.wrappers.scikit_learn', 'tensorflow.contrib.training.python.training.hparam', 'tensorflow.contrib.layers', 'tensorflow.contrib.sparsemax.python.ops', 'fnmatch', 'tensorflow.python.framework.graph_util_impl', 'tensorflow.contrib.learn.python.learn.learn_io', 'tensorflow.python.ops.clip_ops', 'tensorflow.python.ops', 'tensorflow.contrib.keras.api.keras', 'tensorflow.contrib.solvers.python.ops.util', 'd99dc76f3a6d0e4eae2cb9be2387d0b2', 'tensorflow.contrib.memory_stats.ops.gen_memory_stats_ops', 'numpy.ma.core', 'tensorflow.python.framework.errors_impl', 'tensorflow.contrib.specs.python', 'tensorflow.contrib.tensorboard.plugins.trace', 'pkg_resources.extern.packaging', 'tensorflow.contrib.learn.python.learn.graph_actions', 'cv2.detail', 'tensorflow.python.ops.sdca_ops', 'tensorflow.python.client.session', 'tensorflow.contrib.rnn.python.ops.lstm_ops', 'base64', 'cv2.xfeatures2d', 'numpy.matrixlib.defmatrix', 'tensorflow.contrib.layers.python.ops.bucketization_op', 'tensorflow.contrib.keras.api.keras.wrappers', 'tensorflow.contrib.layers.python.layers.summaries', 'tensorflow.python.util.deprecation', 'tensorflow.contrib.slim.python.slim.evaluation', 'google.protobuf.internal.api_implementation', 'matplotlib.fontconfig_pattern', 'cv2.datasets', 'tensorflow.python.ops.nn_impl', 'importlib.machinery', 'tensorflow.python.ops.sets_impl', 'tensorflow.contrib.distributions.python.ops.poisson', 'socket', 'encodings', 'google.protobuf.internal.well_known_types', 'numpy.fft.fftpack', 'cv2.ipp', 'tokenize', 'cv2.ft', 'json.decoder', 'tensorflow.python.estimator.run_config', 'token', 'tensorflow.contrib.linalg.python', 'tensorflow.python.training.ftrl', 'tensorflow.python.ops.data_flow_grad', 'tensorflow.contrib.integrate.python', 'tensorflow.contrib.nn.python.ops', 'tensorflow.contrib.framework.python.framework.checkpoint_utils', 'tensorflow.contrib.keras.python.keras.layers.recurrent', 'tensorflow.contrib.distributions.python.ops.bijectors.sigmoid', 'tensorflow.contrib.quantization.python.array_ops', 'tensorflow.contrib.distributions.python.ops.bijectors.softplus_impl', 'abc', 'zipimport', 'tensorflow.contrib.distributions.python.ops.operator_pd_identity', 'tensorflow.contrib.rnn.python.ops.core_rnn_cell_impl', 'tensorflow.contrib.keras.python.keras.engine', 'tensorflow.contrib.tfprof.python.tools.tfprof', 'tensorflow.python.summary.writer', 'google.protobuf.descriptor_pool', 'tensorflow.contrib.linear_optimizer.python.ops', 'tensorflow.contrib.distributions.python.ops.bijectors.power_transform_impl', 'tensorflow.contrib.quantization.python.nn_ops', 'tensorflow.python.training.training', 'tensorflow.contrib.training.python.training.failure_tolerator', 'tensorflow.python.training.adam', 'tensorflow.core.example.feature_pb2', 'tensorflow.contrib.linalg.python.ops.linear_operator_diag', 'tensorflow.contrib.distributions.python.ops.deterministic', 'tensorflow.core.protobuf.saver_pb2', 'tensorflow.contrib.learn.python.learn.utils.input_fn_utils', 'tensorflow.contrib.keras.python.keras.utils.layer_utils', 'tensorflow.contrib.learn.python.learn.dataframe.dataframe', 'tensorflow.python.ops.embedding_ops', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.difference', 'tensorflow.contrib.training.python', 'cv2.bioinspired', 'logging', 'subprocess', 'tensorflow.python.util.nest', 'cv2.ximgproc.segmentation', 'tensorflow.python.ops.partitioned_variables', 'tensorflow.contrib.learn.python.learn.preprocessing.categorical_vocabulary', 'numpy.lib.nanfunctions', 'tensorflow.contrib.keras.python.keras.datasets.cifar10', 'datetime', 'tensorflow.contrib.distributions.python.ops.multinomial', 'utils.utils_tool', 'urllib', '_bz2', '_compression', 'encodings.utf_8', 'tensorflow.contrib.distributions.python.ops.bijectors.inline_impl', 'tensorflow.contrib.sparsemax.python', 'numpy.polynomial.polynomial', 'uu', 'tensorflow.contrib.labeled_tensor', 'tensorflow.python.client.timeline', 'tensorflow.core.framework.graph_pb2', 'tensorflow.contrib.linalg.python.ops.linear_operator', 'google.protobuf.symbol_database', 'tensorflow.contrib.losses', 'tensorflow.contrib.tensor_forest.python', 'six.moves', 'tensorflow.core.framework.tensor_pb2', 'numpy.core._add_newdocs', 'tensorflow.contrib.learn.python.learn.experiment', 'cv2.plot', 'tensorflow.contrib.slim.python.slim.data.dataset', 'tensorflow.python.estimator.estimator_lib', 'tensorflow.contrib.distributions.python.ops.operator_pd_vdvt_update', 'tensorflow.python.lib', 'tensorflow.contrib.graph_editor.edit', 'tensorflow.python.layers', 'tensorflow.contrib.learn.python.learn.learn_io.numpy_io', 'tensorflow.python.ops.nn', 'tensorflow.contrib.learn.python.learn.ops.embeddings_ops', 'tensorflow.contrib.keras.python.keras.losses', 'tensorflow.python.ops.gradients', 'tensorflow.contrib.learn.python.learn.export_strategy', 'tensorflow.python.ops.control_flow_grad', 'tensorflow.python.training.server_lib', 'tensorflow.contrib.solvers', 'tensorflow.contrib.integrate.python.ops.odes', 'tensorflow.contrib.layers.python.layers.embedding_ops', 'numpy.polynomial.hermite_e', 'tensorflow.contrib.tensor_forest.python.ops.data_ops', 'tensorflow.python.framework.common_shapes', 'tensorflow.contrib.layers.python.layers.utils', 'numpy.lib.index_tricks', 'tensorflow.contrib.learn.python.learn.dataframe.series', 'numpy.linalg.info', 'tensorflow.python.ops.io_ops', 'tensorflow.contrib.keras.python.keras.utils.conv_utils', 'tensorflow.python.framework.op_def_library', 'tensorflow.contrib.cloud.python.ops.gen_bigquery_reader_ops', 'tensorflow.python.ops.gen_resource_variable_ops', 'tensorflow.contrib.keras.python.keras.constraints', 'a0c893eca6b687fe52f72782f634c697', 'tensorflow.contrib.linalg.python.ops.linear_operator_addition', 'tensorflow.python.ops.gen_functional_ops', 'tensorflow.contrib.labeled_tensor.python.ops.ops', 'tensorflow.python.summary.plugin_asset', 'pkg_resources.extern.pyparsing', 'tensorflow.contrib.slim.python.slim.queues', 'hashlib', 'ast', 'numpy.linalg', 'tensorflow.contrib.distributions.python.ops.operator_pd', 'tensorflow.contrib.specs.python.specs_lib', 'tensorflow.contrib.keras.python.keras.models', 'urllib.error', 'cv2.Error', 'tensorflow.contrib.metrics.python.ops.confusion_matrix_ops', 'tensorflow.contrib.integrate', 'tensorflow.contrib.image.python.ops', 'tensorflow.python.platform.benchmark', 'tensorflow.contrib.crf.python', 'tensorflow.python.layers.utils', 'tensorflow.contrib.tensorboard.plugins.trace.trace_info_pb2', 'tensorflow.python.summary', 'cv2.dnn', 'tensorflow.contrib.learn.python.learn.estimators._sklearn', 'tensorflow.contrib.distributions.python.ops.bijectors.bijector', 'numpy', '21de8d59637e674d4b2e45e5064942a0', 'tensorflow.python.ops.script_ops', 'tensorflow.contrib.keras.python.keras.layers.normalization', 'tensorflow.contrib.factorization.python.ops.gen_clustering_ops', 'tensorflow.contrib.memory_stats', 'email.message', 'numpy.linalg.linalg', 'matplotlib', 'tensorflow.contrib.linear_optimizer', 'pkg_resources.extern.packaging.markers', 'tensorflow.contrib.layers.python', 'tensorflow.contrib.rnn.python.ops.fused_rnn_cell', 'tensorflow.contrib.learn.python.learn.dataframe', 'tensorflow.contrib.training.python.training', 'numpy.lib.info', 'tensorflow.contrib.image.python.ops.image_ops', 'tensorflow.core.protobuf.meta_graph_pb2', 'heapq', 'tensorflow.contrib.framework.python.ops.variables', 'csv', 'tensorflow.python.estimator', 'tensorflow.core.lib.core', 'tensorflow.contrib.learn.python.learn.utils.saved_model_export_utils', 'numpy.lib.scimath', 'tensorflow.contrib.metrics.python.ops', 'utils', 'tensorflow.python.ops.gen_linalg_ops', '_ssl', 'struct', 'numpy.core._methods', 'tensorflow.contrib.bayesflow.python.ops.stochastic_tensor', 'tensorflow.contrib.image.python', 'tensorflow.python.estimator.inputs', 'io', 'tensorflow.python.estimator.export', 'tensorflow.contrib.opt.python.training', 'tensorflow.contrib.distributions.python.ops.mixture', 'tensorflow.python.ops.losses.util', 'cv2.aruco', 'pyparsing', 'numpy.fft.info', 'tensorflow.contrib.distributions.python.ops', 'collections.abc', 'posixpath', 'json.encoder', 'pkg_resources.extern.packaging.version', 'tensorflow.contrib.keras.python', 'tensorflow.contrib.layers.python.layers.initializers', 'tensorflow.contrib.layers.python.layers.encoders', 'tensorflow.contrib.bayesflow', 'tensorflow.contrib.keras.api.keras.constraints', 'tensorflow.python.layers.base', 'tensorflow.contrib.legacy_seq2seq.python.ops.seq2seq', 'google.protobuf.pyext', 'tensorflow.core.framework.tensor_description_pb2', 'tensorflow.python.estimator.model_fn', 'tensorflow.python.ops.gen_state_ops', 'tensorflow.contrib.learn.python.learn.models', '_imp', 'tensorflow.contrib.rnn.python.ops.rnn_cell', 'tensorflow.contrib.distributions.python.ops.geometric', 'tensorflow.tools.tfprof.tfprof_log_pb2', 'google.protobuf.json_format', 'tensorflow.python.ops.gen_control_flow_ops', 'tensorflow.python.framework.device', 'tensorflow.contrib.distributions.python.ops.operator_pd_diag', 'tensorflow.contrib.distributions.python.ops.bijectors', 'marshal', 'tensorflow.contrib.learn.python.learn.estimators.model_fn', 'importlib._bootstrap', 'tensorflow.python.user_ops', 'ntpath', 'cv2.optflow', 'uuid', 'xml', 'ipaddress', 'tensorflow.contrib.distributions.python.ops.kullback_leibler', 'tensorflow.python.estimator.inputs.queues', 'tensorflow.contrib.learn.python.learn.preprocessing', 'tensorflow.python.platform.test', 'tensorflow.python.ops.spectral_grad', 'numpy.lib.function_base', 'tensorflow.contrib.distributions.python.ops.logistic', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.csv_parser', 'tensorflow.contrib.keras.python.keras.layers.convolutional', 'email._policybase', 'tensorflow.contrib.keras.python.keras.preprocessing.text', 'numpy.lib.polynomial', '_weakref', 'importlib', '_frozen_importlib_external', 'tensorflow.python.ops.gen_candidate_sampling_ops', 'cv2.reg', 'tensorflow.contrib.rnn.python.ops.rnn', 'tensorflow.python.platform.app', 'tensorflow.contrib.labeled_tensor.python.ops.io_ops', 'tensorflow.tools.tfprof.tfprof_options_pb2', 'tensorflow.python.training.sync_replicas_optimizer', 're', 'dateutil', 'platform', 'tensorflow.python.ops.gradients_impl', 'tensorflow.python.ops.numerics', '_ctypes', 'tensorflow.python.user_ops.user_ops', 'tensorflow.contrib.learn.python.learn.trainable', 'tensorflow.contrib.keras.python.keras.wrappers', 'tensorflow.contrib.keras.python.keras.applications.vgg19', 'email.feedparser', 'numpy.random.mtrand', 'tensorflow.contrib.keras', 'tensorflow.python.saved_model.utils_impl', 'cv2.linemod', 'cv2.text', 'tensorflow.contrib.input_pipeline', 'tensorflow.contrib.keras.api.keras.metrics', 'tensorflow.contrib.ndlstm.python.lstm2d', 'tensorflow.python.training.training_ops', 'tensorflow.python.ops.parsing_ops', 'json.scanner', 'zlib', 'tensorflow.contrib.keras.python.keras.datasets.reuters', 'tensorflow.contrib.distributions.python.ops.bijectors.exp', 'tensorflow.contrib.graph_editor.subgraph', 'tensorflow.python.ops.image_ops_impl', 'lzma', 'tensorflow.contrib.slim.python.slim.data', 'tensorflow.contrib.slim.python.slim.data.dataset_data_provider', 'tensorflow.contrib.distributions.python.ops.mvn_tril', 'tempfile', 'tensorflow.contrib.nn.python.ops.cross_entropy', '_heapq', 'tensorflow.contrib.legacy_seq2seq.python', 'tensorflow.contrib.grid_rnn', 'tensorflow.python.saved_model.loader', 'tensorflow.python.training.rmsprop', 'xml.parsers', 'tensorflow.contrib.keras.python.keras.preprocessing.sequence', 'tensorflow.python.ops.standard_ops', 'numpy.lib.arraypad', 'gc', 'tensorflow.python.training.gen_training_ops', 'ctypes', 'tensorflow.contrib.opt.python.training.external_optimizer', 'google.protobuf.message_factory', 'tensorflow.python.util.protobuf', 'encodings.aliases', 'mtrand', 'tensorflow.contrib.keras.python.keras.initializers', 'tensorflow.core.framework.attr_value_pb2', 'tensorflow.contrib.keras.api.keras.datasets.cifar10', 'tensorflow.contrib.learn.python.learn.datasets.base', 'tensorflow.contrib.crf', 'tensorflow.python.framework.constant_op', 'tensorflow.python.framework.load_library', 'tensorflow.contrib.testing', 'google.protobuf.descriptor_database', 'numpy.testing._private.utils', 'tensorflow.contrib.cudnn_rnn.python.ops.cudnn_rnn_ops', 'tensorflow.python.ops.session_ops', 'calendar', 'google.protobuf.any_pb2', 'tensorflow.python.summary.writer.writer_cache', 'tensorflow.contrib.learn.python.learn.dataframe.queues.feeding_functions', 'tensorflow.contrib.keras.python.keras.applications.inception_v3', 'numpy.ma.extras', 'xml.parsers.expat', 'tensorflow.python.training.supervisor', '_pywrap_tensorflow_print_model_analysis_lib', '_socket', 'numpy.core', 'tensorflow.contrib.session_bundle.exporter', 'zipfile', 'tensorflow.contrib.slim.python.slim.data.data_provider', 'numpy.core._multiarray_tests', 'numpy.core.arrayprint', 'tensorflow.python.training.optimizer', 'tensorflow.python.framework.ops', 'tensorflow.contrib.learn.python.learn.datasets', 'numpy.testing._private.nosetester', 'os.path', 'urllib.parse', 'bisect', 'tensorflow.contrib.losses.python.losses.loss_ops', '_collections', 'tensorflow.contrib.linear_optimizer.python.sdca_optimizer', 'tensorflow.contrib.slim.python.slim', 'tensorflow.contrib.learn.python.learn.utils', 'tensorflow.contrib.rnn.python.ops.core_rnn', 'future', 'numpy.linalg._umath_linalg', 'pwd', 'tarfile', 'tensorflow.contrib.distributions.python.ops.distribution_util', 'tensorflow.contrib.solvers.python.ops.lanczos', 'urllib.request', 'tensorflow.python.ops.gen_nn_ops', 'tensorflow.python.saved_model.signature_def_utils', 'numpy.testing._private', 'tensorflow.contrib.cloud.python.ops', 'tensorflow.python.training.basic_session_run_hooks', 'numpy.testing._private.decorators', 'tensorflow.contrib.keras.api.keras.datasets.boston_housing', 'tensorflow.contrib.training', 'tensorflow.contrib.learn.python.learn.estimators.metric_key', 'tensorflow.contrib.keras.python.keras.applications.imagenet_utils', 'tensorflow.contrib.grid_rnn.python', 'pkg_resources.extern.packaging.requirements', 'tensorflow.core.framework.function_pb2', 'tensorflow.contrib.training.python.training.resample', 'tensorflow.contrib.slim.python.slim.model_analyzer', 'signal', 'tensorflow.python.layers.core', 'nets.resnet', 'xml.parsers.expat.model', 'pyexpat.model', 'tensorflow.python.pywrap_tensorflow', 'tensorflow.contrib.distributions.python.ops.mvn_diag_plus_low_rank', 'tensorflow.contrib.keras.python.keras.utils.np_utils', 'math', '_locale', 'weakref', 'tensorflow.python.framework.tensor_shape', 'tensorflow.contrib.saved_model.python', 'tensorflow.contrib.distributions.python.ops.relaxed_bernoulli', 'difflib', 'tensorflow.contrib.distributions.python.ops.chi2', 'tensorflow.python.framework', 'codecs', 'tensorflow.contrib.session_bundle.manifest_pb2', 'tensorflow.contrib.ndlstm', 'pkg_resources._vendor.six', 'tensorflow.contrib.distributions.python.ops.bijectors.cholesky_outer_product', 'pkg_resources.extern.six.moves', 'numpy.fft.helper', 'tensorflow.python.ops.metrics_impl', 'tensorflow.python.ops.array_ops', 'copy', 'tensorflow.contrib.learn.python.learn.session_run_hook', 'tensorflow.contrib.distributions.python.ops.negative_binomial', 'tensorflow.python.training.device_setter', 'tensorflow.python.estimator.inputs.pandas_io', 'keyword', 'google', 'tensorflow.contrib.tensor_forest', 'pathlib', 'email._encoded_words', 'tensorflow.python.util.compat', 'shutil', 'tensorflow.contrib.tensor_forest.python.ops.tensor_forest_ops', 'tensorflow.python.lib.io.file_io', 'tensorflow.python.training.gradient_descent', '6f42c22f2523ba04a92beabc6382ea6a', 'tensorflow.contrib.learn.python.learn', 'tensorflow.contrib.keras.api.keras.datasets.cifar100', 'tensorflow.python.ops.gen_random_ops', 'swig_runtime_data4', 'tensorflow.python.estimator.inputs.queues.feeding_functions', 'tensorflow.core.framework.step_stats_pb2', 'tensorflow.contrib.distributions.python.ops.bijectors.sigmoid_centered', 'tensorflow.contrib.distributions.python.ops.relaxed_onehot_categorical', 'tensorflow.python.ops.gen_math_ops', 'tensorflow.python.ops.gen_string_ops', 'tensorflow.contrib.learn.python.learn.utils.gc', 'unittest.suite', 'tensorflow.core.example', 'tensorflow.contrib.compiler.jit', 'encodings.latin_1', 'tensorflow.contrib.testing.python.framework', 'argparse', 'cv2.multicalib', 'itertools', 'tensorflow.core.protobuf.rewriter_config_pb2', 'tensorflow.python.ops.resource_variable_ops', 'pkg_resources.extern.packaging._compat', 'tensorflow.contrib.keras.api.keras.applications.resnet50', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.hashes', 'tensorflow.contrib.keras.api.keras.backend', 'de01d2588a57b3f7d752d74e2796a249', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.binary_transforms', 'numpy.polynomial._polybase', 'tensorflow.contrib.framework.python.ops.ops', 'cv2', 'tensorflow.contrib.learn.python.learn.estimators.tensor_signature', 'numpy.lib.twodim_base', 'importlib.util', 'tensorflow.python.saved_model.signature_def_utils_impl', 'tensorflow.contrib.ndlstm.python', 'tensorflow.python.platform.tf_logging', 'tensorflow.contrib.keras.api.keras.utils', 'tensorflow.contrib.copy_graph', 'tensorflow.contrib.slim.python.slim.learning', 'tensorflow.contrib.distributions.python.ops.dirichlet_multinomial', 'tensorflow.python.ops.tensor_array_grad', 'tensorflow.contrib.layers.python.layers.feature_column_ops', 'tensorflow.contrib.distributions.python.ops.bijectors.chain_impl', 'tensorflow.contrib.copy_graph.python', 'os', 'tensorflow.python.ops.spectral_ops', 'tensorflow.contrib.distributions.python.ops.wishart', 'tensorflow.contrib.layers.ops.gen_sparse_feature_cross_op', 'tensorflow.contrib.layers.ops.gen_bucketization_op', 'tensorflow.contrib.ndlstm.python.misc', 'tensorflow.contrib.keras.api', 'tensorflow.contrib.tfprof.python.tools.tfprof.tfprof_logger', 'cv2.data', 'matplotlib.rcsetup', 'tensorflow.contrib.cudnn_rnn', 'tensorflow.contrib.opt', 'tensorflow.python.framework.sparse_tensor', 'tensorflow.contrib.grid_rnn.python.ops', 'tensorflow.contrib.bayesflow.python.ops', 'tensorflow.contrib.learn.python.learn.summary_writer_cache', 'tensorflow.contrib.layers.python.layers.target_column', 'tensorflow.contrib.distributions.python.ops.uniform', 'tensorflow.python.estimator.export.export', 'tensorflow.python.lib.io.python_io', 'tensorflow.contrib.sparsemax.python.ops.sparsemax_loss', 'tensorflow.contrib.distributions.python.ops.bijectors.cholesky_outer_product_impl', '_warnings', 'tensorflow.python.client', 'tensorflow.python.framework.function', 'tensorflow.python.ops.gradient_checker', 'google.protobuf.descriptor_pb2', 'tensorflow.contrib.rnn.python.ops.gru_ops', 'tensorflow.contrib.legacy_seq2seq.python.ops', 'tensorflow.contrib.distributions.python.ops.transformed_distribution', 'tensorflow.contrib.keras.python.keras.applications.resnet50', 'tensorflow.contrib.tensor_forest.python.ops.gen_tensor_forest_ops', 'pyexpat.errors', 'tensorflow.contrib.keras.python.keras.activations', 'pkg_resources._vendor', 'tensorflow.contrib.keras.api.keras.losses', 'unittest.util', 'tensorflow.contrib.slim.python.slim.data.data_decoder', 'pkg_resources._vendor.six.moves', '_compat_pickle', 'cv2.flann', 'tensorflow.contrib.copy_graph.python.util.copy_elements', 'grp', 'urllib.response', 'tensorflow.python.ops.data_flow_ops', 'stat', 'numpy.lib.npyio', 'tensorflow.contrib.training.python.training.sampling_ops', 'utils.data_provider.data_provider', 'tensorflow.contrib.crf.python.ops.crf', 'tensorflow.contrib.keras.api.keras.preprocessing.sequence', 'tensorflow.contrib.framework.python.ops.gen_variable_ops', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.example_parser', 'tensorflow.core.protobuf.queue_runner_pb2', 'google.protobuf.internal.encoder', 'bz2', 'google.protobuf.descriptor', '_operator', 'tensorflow.contrib.learn.python.learn.ops', 'tensorflow.contrib.slim.python.slim.data.prefetch_queue', 'tensorflow.contrib.rnn.ops.gen_lstm_ops', 'tensorflow.core.framework.resource_handle_pb2', 'sys', 'tensorflow.contrib.seq2seq.python.ops.loss', 'email.errors', 'tensorflow.contrib.labeled_tensor.python.ops', 'tensorflow.python.training.input', 'tensorflow.contrib.graph_editor.select', 'numpy.core.function_base', 'numpy.core.einsumfunc', '_csv', 'tensorflow.contrib.framework', 'tensorflow.contrib.framework.python.framework.tensor_util', 'selectors', 'six.moves.urllib', 'pyexpat', 'numpy.core.getlimits', 'tensorflow.core.framework.allocation_description_pb2', 'cv2.ocl', 'cv2.omnidir', 'numpy.random', 'tensorflow.contrib.distributions.python.ops.bijectors.identity_impl', 'tensorflow.contrib.distributions.python.ops.bijectors.invert', 'tensorflow.contrib.framework.python.ops', 'tensorflow.contrib.image.ops', 'tensorflow.contrib.linalg.python.ops.linear_operator_tril', 'tensorflow.contrib.graph_editor.util', 'time', 'tensorflow.contrib.input_pipeline.ops', 'tensorflow.contrib.labeled_tensor.python.ops._typecheck', 'tensorflow.python.saved_model', 'tensorflow.python.ops.weights_broadcast_ops', 'tensorflow.contrib.factorization.python.ops.factorization_ops', 'tensorflow.contrib.layers.python.layers.optimizers', 'tensorflow.contrib.distributions.python.ops.mvn_diag', 'tensorflow.contrib.tensorboard.plugins.trace.trace', '_codecs', 'glob', 'tensorflow.contrib.learn.python.learn.datasets.text_datasets', 'tensorflow.contrib.learn.python.learn.learn_io.data_feeder', 'tensorflow.python.ops.sparse_grad', 'tensorflow.contrib.distributions.python.ops.bijectors.inline', 'tensorflow.contrib.cudnn_rnn.ops', 'pkg_resources.extern', 'tensorflow.python.ops.resources', 'tensorflow.contrib.distributions.python.ops.bijectors.chain', 'tensorflow.contrib.distributions.python.ops.bijectors.softmax_centered', 'tensorflow.contrib.learn.python.learn.estimators.estimator', 'tensorflow.python.training.saver', 'numpy._distributor_init', 'tensorflow.contrib.keras.python.keras.layers.embeddings', 'tensorflow.contrib.distributions.python.ops.special_math', 'tensorflow.contrib.losses.python.losses', 'multiprocessing.process', 'tensorflow.contrib.factorization.python', 'tensorflow.contrib.metrics.python.ops.set_ops', 'tensorflow.contrib.tfprof.python.tools', 'types', 'tensorflow.python.client.client_lib', 'tensorflow.contrib.quantization.python', 'tensorflow.contrib.bayesflow.python.ops.variational_inference', 'tensorflow.contrib.tensor_forest.python.tensor_forest', 'tensorflow.python.ops.string_ops', 'cv2.cuda', 'nets.resnet.resnet_v1', '_ast', 'cython_runtime', 'tensorflow.contrib.compiler', 'tensorflow.contrib.saved_model.python.saved_model.signature_def_utils', 'tensorflow.contrib.framework.python.ops.arg_scope', 'tensorflow.python.ops.init_ops', 'tensorflow.python.ops.linalg_ops', 'dis', 'google.protobuf', 'numpy.lib._datasource', 'google.protobuf.text_encoding', 'tensorflow.python.ops.image_grad', 'tensorflow.contrib.distributions.python.ops.bijectors.bijector_impl', 'matplotlib._color_data', 'tensorflow.contrib.distributions.python.ops.operator_pd_cholesky', 'tensorflow.core.protobuf.config_pb2', 'tensorflow.contrib.learn.python.learn.learn_io.graph_io', 'tensorflow.contrib.distributions.python.ops.inverse_gamma', 'tensorflow.contrib.distributions.python.ops.conditional_transformed_distribution', 'tensorflow.contrib.bayesflow.python.ops.stochastic_variables', 'tensorflow.contrib.learn.python.learn.estimators.dnn_linear_combined', 'tensorflow.contrib.distributions.python.ops.bijectors.affine_linear_operator_impl', 'tensorflow.contrib.losses.python', 'binascii', 'tensorflow.contrib.bayesflow.python.ops.stochastic_graph', 'tensorflow.contrib.keras.python.keras.utils.io_utils', 'tensorflow.python.training.session_run_hook', 'tensorflow.python.summary.summary', 'tensorflow.contrib.specs.python.specs', 'collections', 'tensorflow.python.ops.losses.losses_impl', 'tensorflow.contrib.keras.python.keras.datasets.mnist', 'cv2.cv2', 'tensorflow.contrib.distributions.python.ops.normal_conjugate_posteriors', 'tensorflow.python.ops.check_ops', 'tensorflow.python.platform.sysconfig', 'tensorflow.python.saved_model.builder_impl', 'tensorflow.python.ops.gen_io_ops', 'dateutil._version', 'numpy.core.fromnumeric', 'tensorflow.contrib.keras.api.keras.applications.vgg16', 'tensorflow.python.ops.special_math_ops', 'cv2.videostab', 'numpy.core._string_helpers', 'tensorflow.contrib.learn.python.learn.estimators.prediction_key', 'numpy.polynomial.chebyshev', 'importlib.abc', 'tensorflow.contrib.learn.python.learn.dataframe.transforms', 'tensorflow.python.lib.io', 'tensorflow.python.training.evaluation', 'tensorflow.contrib.bayesflow.python.ops.variational_inference_impl', 'tensorflow.python.util', 'tensorflow.core.util.test_log_pb2', 'tensorflow.contrib.learn.python', 'tensorflow.contrib.distributions.python.ops.gamma', 'tensorflow.contrib.sparsemax.python.ops.sparsemax', 'tensorflow.contrib.solvers.python.ops', 'numbers', 'tensorflow.tools.tfprof.tfprof_output_pb2', 'tensorflow.python.training.momentum', 'numpy.lib._version', 'cv2.fisheye', 'tensorflow.contrib.linalg', 'tensorflow.python.ops.histogram_ops', 'tensorflow.python.estimator.inputs.inputs', 'tensorflow.contrib.keras.python.keras.utils', 'tensorflow.contrib.linear_optimizer.python.ops.sharded_mutable_dense_hashtable', 'tensorflow.contrib.image.ops.gen_image_ops', 'tensorflow.contrib.slim.python.slim.data.parallel_reader', 'tensorflow.python.saved_model.saved_model', 'nets', 'google.protobuf.internal.wire_format', 'tensorflow.python.util.protobuf.compare', 'numpy.core.overrides', 'tensorflow.contrib.keras.python.keras.datasets.cifar', 'tensorflow.python.framework.errors', 'numpy.compat.py3k', 'cv2.ml', 'tensorflow.contrib.framework.python', 'numpy.polynomial.legendre', 'tensorflow.contrib.distributions.python.ops.beta', 'email.quoprimime', 'importlib._bootstrap_external', 'tensorflow.python.framework.tensor_util', 'numpy.core.multiarray', 'tensorflow.contrib.specs.python.params_ops', 'tensorflow.tools', 'tensorflow.python.platform.resource_loader', '_stat', 'numpy.lib.type_check', 'six.moves.urllib.error', 'unittest', 'tensorflow.contrib.training.python.training.tuner', 'tensorflow.contrib.deprecated', 'tensorflow.python.ops.candidate_sampling_ops', '_struct', 'tensorflow.python.ops.losses', 'tensorflow.contrib.keras.api.keras.wrappers.scikit_learn', 'sysconfig', 'tensorflow.python.saved_model.constants', 'tensorflow.python.ops.gen_user_ops', 'tensorflow.contrib.distributions.python.ops.quantized_distribution', 'tensorflow.contrib.keras.python.keras.engine.topology', 'numpy.core._internal', 'tensorflow.contrib.learn.python.learn.estimators.constants', 'numpy.core.machar', 'tensorflow.python.ops.tensor_array_ops', 'numpy._pytesttester', 'tensorflow.contrib.input_pipeline.python', 'tensorflow.contrib.keras.python.keras.datasets', 'tensorflow.contrib.distributions.python.ops.bijectors.softplus', 'tensorflow.contrib.keras.api.keras.models', 'cv2.bgsegm', 'tensorflow.contrib.keras.api.keras.applications.inception_v3', 'tensorflow.python.pywrap_tensorflow_internal', 'tensorflow.contrib.saved_model', 'tensorflow.contrib.keras.api.keras.layers', 'tensorflow.contrib.memory_stats.python', 'tensorflow.contrib.linear_optimizer.python.ops.sdca_ops', 'tensorflow.contrib.learn.python.learn.evaluable', 'tensorflow.contrib.factorization', 'tensorflow.python.framework.graph_io', 'tensorflow.contrib.distributions.python.ops.bijectors.affine_linear_operator', 'tensorflow.contrib.learn.python.learn.estimators.svm', 'numpy.lib.ufunclike', '_frozen_importlib', 'tensorflow.contrib.learn.python.learn.learn_io.dask_io', 'tensorflow.contrib.learn.python.learn.monitors', 'tensorflow.contrib.learn.python.learn.dataframe.queues', 'tensorflow.contrib.layers.python.ops', 'tensorflow.contrib.keras.api.keras.datasets.imdb', 'quopri', 'tensorflow.contrib.crf.python.ops', 'tensorflow.python.ops.nn_grad', 'tensorflow.python.estimator.export.export_lib', 'tensorflow.contrib.distributions.python.ops.bernoulli', 'tensorflow.contrib.metrics.python.metrics', 'encodings.raw_unicode_escape', 'tensorflow.contrib.tfprof.python', 'tensorflow.python.summary.summary_iterator', 'tensorflow.contrib.linalg.python.ops.linear_operator_util', 'numpy.fft', 'tensorflow.contrib.bayesflow.python', 'tensorflow.python.training.queue_runner', 'tensorflow.contrib.tfprof.python.tools.tfprof.pywrap_tensorflow_print_model_analysis_lib', 'tensorflow.contrib.learn.python.learn.basic_session_run_hooks', 'tensorflow.contrib.learn.python.learn.estimators.run_config', 'unittest.runner', 'contextlib', 'pkg_resources', 'tensorflow.python.framework.graph_util', 'gzip', 'tensorflow.contrib.keras.python.keras.optimizers', 'numpy.lib.arraysetops', 'tensorflow.contrib.distributions.python.ops.categorical', 'enum', 'tensorflow.python.ops.gen_logging_ops', 'tensorflow.contrib.distributions.python.ops.bijectors.exp_impl', 'pickle', 'tensorflow.python.ops.summary_ops', 'tensorflow.contrib.layers.python.ops.sparse_ops', 'tensorflow.contrib', 'tensorflow.contrib.util', 'tensorflow.contrib.bayesflow.python.ops.entropy_impl', 'cv2.saliency', 'mpl_toolkits', 'tensorflow.contrib.distributions.python.ops.bijectors.sigmoid_centered_impl', 'tensorflow.contrib.rnn.ops.gen_gru_ops', 'gettext', 'queue', 'tensorflow.python.ops.state_grad', 'tensorflow.contrib.tensor_forest.client.random_forest', 'tensorflow.python.estimator.estimator', 'tensorflow.python.ops.linalg_grad', 'tensorflow.contrib.layers.python.layers', 'tensorflow.python.training.checkpoint_state_pb2', 'tensorflow.contrib.tfprof.python.tools.tfprof.model_analyzer', '_pywrap_tensorflow_internal', 'tensorflow.contrib.learn.python.learn.datasets.mnist', 'numpy.lib.format', 'tensorflow.contrib.labeled_tensor.python', 'tensorflow.contrib.cloud', 'http.client', 'tensorflow.contrib.keras.python.keras.datasets.cifar100', 'tensorflow.python.platform.gfile', 'tensorflow.python.training.moving_averages', 'tensorflow.contrib.bayesflow.python.ops.stochastic_graph_impl', 'tensorflow.core.protobuf.tensorflow_server_pb2', 'tensorflow.contrib.keras.api.keras.applications.vgg19', '_opcode', 'tensorflow.contrib.distributions.python.ops.bijectors.identity', 'tensorflow.contrib.ndlstm.python.lstm1d', 'tensorflow.contrib.learn.python.learn.ops.seq2seq_ops', 'tensorflow.contrib.training.python.training.evaluation', 'email.base64mime', 'tensorflow.contrib.keras.python.keras.layers.noise', 'tensorflow.contrib.training.python.training.hparam_pb2', 'tensorflow.python.ops.state_ops', 'random', 'tensorflow.python.lib.io.tf_record', 'numpy.lib', 'tensorflow.python.estimator.inputs.numpy_io', 'google.protobuf.pyext._message', '_thread', 'cycler', 'tensorflow.contrib.input_pipeline.ops.gen_input_pipeline_ops', 'tensorflow.core.lib', 'tensorflow.python.ops.gen_sparse_ops', 'tensorflow.contrib.keras.python.keras.datasets.boston_housing', 'c25bf247d6a4a71428ea400372969416', 'tensorflow.python.ops.image_ops', 'tensorflow.python.estimator.inputs.queues.feeding_queue_runner', 'tensorflow.contrib.cloud.python.ops.bigquery_reader_ops', '_posixsubprocess', 'tensorflow.contrib.distributions.python.ops.distribution', 'tensorflow.contrib.keras.api.keras.activations', 'tensorflow.python.framework.test_util', 'tensorflow.core.framework.node_def_pb2', 'tensorflow.python.framework.importer', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.in_memory_source', 'matplotlib._version', 'tensorflow.python', 'tensorflow.contrib.keras.python.keras.layers', 'tensorflow.contrib.seq2seq', '_sre', 'plistlib', 'tensorflow.python.ops.logging_ops', 'tensorflow.contrib.factorization.python.ops.gmm', 'pkg_resources._vendor.packaging.about', 'tensorflow.contrib.keras.python.keras.layers.core', 'tensorflow.python.ops.sparse_ops', 'tensorflow.contrib.learn.python.learn.dataframe.transform', 'numpy.core.umath', 'tensorflow.contrib.framework.python.ops.prettyprint_ops', 'tensorflow.python.ops.rnn_cell_impl', 'tensorflow.contrib.keras.api.keras.preprocessing.text', 'distutils', 'tensorflow.contrib.keras.python.keras.layers.local', 'google.protobuf.message', 'tensorflow.contrib.learn.python.learn.estimators.head', 'numpy.ctypeslib', 'email.charset', 'tensorflow.python.ops.variables', 'tensorflow.contrib.tensorboard.plugins', 'google.protobuf.internal.decoder', '78da087da6dda04802f331e8197a652c', 'numpy.core._dtype_ctypes', 'numpy.matrixlib', 'six', 'tensorflow.contrib.nn.python', 'tensorflow.contrib.metrics.python.ops.histogram_ops', 'tensorflow.contrib.learn', 'tensorflow.contrib.learn.python.learn.estimators.linear', 'tensorflow.contrib.graph_editor.reroute', 'json', 'tensorflow.core.framework.device_attributes_pb2', 'tensorflow.python.ops.sets', 'tensorflow.python.training.proximal_adagrad', 'tensorflow.contrib.nn', 'tensorflow.python.ops.control_flow_ops', 'pkg_resources.extern.six', 'traceback', 'tensorflow.python.util.all_util', 'tensorflow.contrib.distributions.python.ops.bijectors.conditional_bijector', 'tensorflow.contrib.keras.python.keras.applications.vgg16', 'tensorflow.contrib.distributions.python.ops.exponential', 'cv2.line_descriptor', 'tensorflow.contrib.training.python.training.feeder', 'utils.data_provider', 'tensorflow.contrib.labeled_tensor.python.ops.core', 'copyreg', 'http', 'numpy.polynomial.laguerre', 'parser', 'unittest.case', 'tensorflow.python.ops.gen_sdca_ops', '_decimal', 'tensorflow.python.framework.registry', 'numpy._globals', 'pkg_resources.py31compat', 'tensorflow.contrib.layers.python.layers.layers', 'tensorflow.python.framework.dtypes', 'tensorflow.contrib.specs.python.specs_ops', 'tensorflow.contrib.tensor_forest.client', 'tensorflow.contrib.graph_editor.transform', 'tensorflow.contrib.rnn.python', 'tensorflow.contrib.bayesflow.python.ops.monte_carlo_impl', 'tensorflow.contrib.seq2seq.python.ops', 'tensorflow.python.ops.rnn', 'tensorflow.contrib.distributions.python.ops.bijectors.sigmoid_impl', 'tensorflow.contrib.bayesflow.python.ops.stochastic_tensor_impl', 'tensorflow.contrib.metrics', 'tensorflow.contrib.linalg.python.ops.linear_operator_composition', 'tensorflow.python.training.learning_rate_decay', 'tensorflow.contrib.image', 'tensorflow.contrib.memory_stats.python.ops', 'tensorflow.tools.tfprof', 'tensorflow.contrib.layers.ops', 'tensorflow.python.ops.template', 'tensorflow.contrib.bayesflow.python.ops.entropy', 'tensorflow.python.ops.losses.losses', 'tensorflow.python.saved_model.signature_constants', 'tensorflow.contrib.metrics.python', 'tensorflow.core.framework.summary_pb2', 'tensorflow.python.summary.writer.event_file_writer', 'tensorflow.contrib.slim.python.slim.data.tfexample_decoder', 'numpy.lib.utils', 'tensorflow.python.util.decorator_utils', 'numpy.core._dtype', 'numpy.compat._inspect', 'tensorflow.contrib.layers.python.layers.regularizers', 'tensorflow.contrib.keras.api.keras.applications.xception', 'tensorflow.contrib.framework.python.framework', 'tensorflow.contrib.copy_graph.python.util', 'tensorflow.python.ops.gen_array_ops', 'tensorflow.contrib.linear_optimizer.python', 'tensorflow.contrib.distributions.python.ops.bijectors.affine_impl', 'tensorflow.python.layers.convolutional', 'tensorflow.contrib.distributions.python.ops.bijectors.power_transform', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.split_mask', 'ctypes.util', 'tensorflow.contrib.labeled_tensor.python.ops.nn', 'tensorflow.contrib.learn.python.learn.estimators.rnn_common', 'tensorflow.contrib.training.python.training.bucket_ops', 'select', 'xml.parsers.expat.errors', 'numpy.compat', 'tensorflow.contrib.learn.python.learn.metric_spec', 'tensorflow.python.layers.pooling', 'tensorflow.contrib.testing.python.framework.fake_summary_writer', 'tensorflow.contrib.memory_stats.python.ops.memory_stats_ops', 'cv2.ppf_match_3d', 'tensorflow.python.ops.array_grad', 'tensorflow.contrib.keras.python.keras.preprocessing.image', 'tensorflow.core.util.event_pb2', 'tensorflow.python.training', 'tensorflow.contrib.factorization.python.ops.gmm_ops', 'tensorflow.contrib.keras.python.keras.layers.wrappers', 'tensorflow.contrib.learn.python.learn.dataframe.tensorflow_dataframe', 'tensorflow.contrib.learn.python.learn.dataframe.transforms.sum', 'tensorflow.contrib.seq2seq.python.ops.basic_decoder', 'tensorflow.python.training.adadelta', 'tensorflow.contrib.quantization', 'numpy.lib.financial', 'tensorflow.contrib.grid_rnn.python.ops.grid_rnn_cell', 'tensorflow.python.training.queue_runner_impl'] DEBUG:matplotlib:CACHEDIR=/root/.cache/matplotlib DEBUG:matplotlib.font_manager:Using fontManager instance from /root/.cache/matplotlib/fontlist-v300.json DEBUG:matplotlib.pyplot:Loaded backend TkAgg version unknown. resnet_v1_50/block1 (?, ?, ?, 256) resnet_v1_50/block2 (?, ?, ?, 512) resnet_v1_50/block3 (?, ?, ?, 1024) resnet_v1_50/block4 (?, ?, ?, 2048) resnet_v1_50/block1 (?, ?, ?, 256) resnet_v1_50/block2 (?, ?, ?, 512) resnet_v1_50/block3 (?, ?, ?, 1024) resnet_v1_50/block4 (?, ?, ?, 2048) 2019-06-16 02:22:46.345406: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. 2019-06-16 02:22:46.345462: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. 2019-06-16 02:22:46.345472: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations. 2019-06-16 02:22:46.345480: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. 2019-06-16 02:22:46.345488: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations. 2019-06-16 02:22:46.948758: I tensorflow/core/common_runtime/gpu/gpu_device.cc:887] Found device 0 with properties: name: Tesla P4 major: 6 minor: 1 memoryClockRate (GHz) 1.1135 pciBusID 0000:05:00.0 Total memory: 7.43GiB Free memory: 7.31GiB 2019-06-16 02:22:46.948930: W tensorflow/stream_executor/cuda/cuda_driver.cc:485] creating context when one is currently active; existing: 0xcf23ab0 2019-06-16 02:22:47.529396: I tensorflow/core/common_runtime/gpu/gpu_device.cc:887] Found device 1 with properties: name: Tesla P4 major: 6 minor: 1 memoryClockRate (GHz) 1.1135 pciBusID 0000:81:00.0 Total memory: 7.43GiB Free memory: 7.31GiB 2019-06-16 02:22:47.529478: I tensorflow/core/common_runtime/gpu/gpu_device.cc:779] Peer access not supported between device ordinals 0 and 1 2019-06-16 02:22:47.529496: I tensorflow/core/common_runtime/gpu/gpu_device.cc:779] Peer access not supported between device ordinals 1 and 0 2019-06-16 02:22:47.529522: I tensorflow/core/common_runtime/gpu/gpu_device.cc:908] DMA: 0 1 2019-06-16 02:22:47.529530: I tensorflow/core/common_runtime/gpu/gpu_device.cc:918] 0: Y N 2019-06-16 02:22:47.529536: I tensorflow/core/common_runtime/gpu/gpu_device.cc:918] 1: N Y 2019-06-16 02:22:47.529550: I tensorflow/core/common_runtime/gpu/gpu_device.cc:977] Creating TensorFlow device (/gpu:0) -> (device: 0, name: Tesla P4, pci bus id: 0000:05:00.0) 2019-06-16 02:22:47.529558: I tensorflow/core/common_runtime/gpu/gpu_device.cc:977] Creating TensorFlow device (/gpu:1) -> (device: 1, name: Tesla P4, pci bus id: 0000:81:00.0) INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part3/image_2117.jpg INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:8034 training images in ./icdar2017rctw_train_v1.2/train/ INFO:root:before shrink poly area:32.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6929.jpg INFO:root:before shrink poly area:30.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1501.jpg INFO:root:before shrink poly area:150.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7908.jpg INFO:root:before shrink poly area:26.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7643.jpg 2019-06-16 02:23:09.599360: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:247] PoolAllocator: After 1640 get requests, put_count=1574 evicted_count=1000 eviction_rate=0.635324 and unsatisfied allocation rate=0.710976 2019-06-16 02:23:09.599395: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:259] Raising pool_sizelimit from 100 to 110 INFO:root:Step 000000, model loss 0.7630, total loss 1.0419, 1.77 seconds/step, 4.52 examples/second INFO:root:Step 000010, model loss 0.6606, total loss 0.9395, 1.93 seconds/step, 4.14 examples/second 2019-06-16 02:23:31.091535: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:247] PoolAllocator: After 2568 get requests, put_count=2121 evicted_count=1000 eviction_rate=0.471476 and unsatisfied allocation rate=0.57243 2019-06-16 02:23:31.091587: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:259] Raising pool_sizelimit from 256 to 281 INFO:root:Step 000020, model loss 0.8125, total loss 1.0914, 1.00 seconds/step, 8.00 examples/second INFO:root:Step 000030, model loss 0.9919, total loss 1.2708, 0.99 seconds/step, 8.06 examples/second 2019-06-16 02:23:50.217105: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:247] PoolAllocator: After 3852 get requests, put_count=3903 evicted_count=1000 eviction_rate=0.256213 and unsatisfied allocation rate=0.261682 2019-06-16 02:23:50.217166: I tensorflow/core/common_runtime/gpu/pool_allocator.cc:259] Raising pool_sizelimit from 655 to 720 INFO:root:Step 000040, model loss 0.8503, total loss 1.1292, 1.00 seconds/step, 8.00 examples/second INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7948.jpg INFO:root:Step 000050, model loss 0.7931, total loss 1.0720, 0.98 seconds/step, 8.14 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_677.jpg INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_677.jpg INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_686.jpg INFO:root:Step 000060, model loss 0.8387, total loss 1.1175, 0.99 seconds/step, 8.07 examples/second INFO:root:Step 000070, model loss 0.5866, total loss 0.8655, 0.99 seconds/step, 8.12 examples/second INFO:root:Step 000080, model loss 0.4927, total loss 0.7715, 0.99 seconds/step, 8.05 examples/second INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1620.jpg INFO:root:Step 000090, model loss 0.9037, total loss 1.1825, 1.47 seconds/step, 5.45 examples/second INFO:root:before shrink poly area:56.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_5449.jpg INFO:root:Step 000100, model loss 0.4607, total loss 0.7395, 1.69 seconds/step, 4.74 examples/second INFO:root:Step 000110, model loss 0.5811, total loss 0.8599, 1.88 seconds/step, 4.25 examples/second INFO:root:Step 000120, model loss 0.7551, total loss 1.0339, 1.72 seconds/step, 4.64 examples/second INFO:root:Step 000130, model loss 0.8812, total loss 1.1600, 1.72 seconds/step, 4.64 examples/second INFO:root:Step 000140, model loss 0.4375, total loss 0.7163, 2.16 seconds/step, 3.71 examples/second INFO:root:Step 000150, model loss 0.7385, total loss 1.0172, 2.43 seconds/step, 3.30 examples/second INFO:root:before shrink poly area:58.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1006.jpg INFO:root:Step 000160, model loss 0.7277, total loss 1.0064, 2.43 seconds/step, 3.29 examples/second INFO:root:Step 000170, model loss 0.5224, total loss 0.8011, 2.43 seconds/step, 3.29 examples/second INFO:root:Step 000180, model loss 0.9714, total loss 1.2501, 2.43 seconds/step, 3.29 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part3/image_2166.jpg INFO:root:Step 000190, model loss 0.5887, total loss 0.8674, 2.43 seconds/step, 3.29 examples/second INFO:root:before shrink poly area:30.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_4927.jpg INFO:root:Step 000200, model loss 0.6891, total loss 0.9678, 2.45 seconds/step, 3.27 examples/second INFO:root:Step 000210, model loss 0.5563, total loss 0.8350, 2.77 seconds/step, 2.89 examples/second INFO:root:before shrink poly area:26.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_4695.jpg INFO:root:before shrink poly area:46.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_4695.jpg INFO:root:Step 000220, model loss 0.5175, total loss 0.7961, 2.61 seconds/step, 3.06 examples/second INFO:root:Step 000230, model loss 0.5861, total loss 0.8648, 2.70 seconds/step, 2.97 examples/second INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:8.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1937.jpg INFO:root:Step 000240, model loss 0.9837, total loss 1.2623, 2.71 seconds/step, 2.95 examples/second INFO:root:Step 000250, model loss 0.7516, total loss 1.0302, 2.75 seconds/step, 2.91 examples/second INFO:root:Step 000260, model loss 0.4626, total loss 0.7412, 2.68 seconds/step, 2.98 examples/second INFO:root:Step 000270, model loss 0.8013, total loss 1.0799, 2.74 seconds/step, 2.92 examples/second INFO:root:Step 000280, model loss 0.4171, total loss 0.6957, 2.86 seconds/step, 2.79 examples/second INFO:root:Step 000290, model loss 0.5999, total loss 0.8785, 2.80 seconds/step, 2.85 examples/second INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_4795.jpg INFO:root:Step 000300, model loss 0.8744, total loss 1.1530, 2.85 seconds/step, 2.80 examples/second INFO:root:before shrink poly area:30.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6528.jpg INFO:root:before shrink poly area:20.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6528.jpg INFO:root:before shrink poly area:60.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6528.jpg INFO:root:before shrink poly area:32.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6528.jpg INFO:root:Step 000310, model loss 0.4642, total loss 0.7427, 3.16 seconds/step, 2.53 examples/second INFO:root:Step 000320, model loss 0.7711, total loss 1.0497, 2.82 seconds/step, 2.84 examples/second INFO:root:Step 000330, model loss 0.5749, total loss 0.8535, 2.85 seconds/step, 2.81 examples/second INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part3/image_2001.jpg INFO:root:Step 000340, model loss 0.8024, total loss 1.0809, 2.79 seconds/step, 2.86 examples/second INFO:root:Step 000350, model loss 0.8053, total loss 1.0838, 2.85 seconds/step, 2.81 examples/second INFO:root:Step 000360, model loss 0.5403, total loss 0.8189, 2.88 seconds/step, 2.78 examples/second INFO:root:Step 000370, model loss 0.5011, total loss 0.7796, 2.83 seconds/step, 2.83 examples/second INFO:root:Step 000380, model loss 0.4545, total loss 0.7330, 2.90 seconds/step, 2.76 examples/second

XiaoLaoDi commented 5 years ago

after some steps training, the log goes to :

XiaoLaoDi commented 5 years ago

INFO:root:Step 011920, model loss 0.5708, total loss 0.8458, 2.00 seconds/step, 3.99 examples/second INFO:root:Step 011930, model loss 0.4588, total loss 0.7339, 1.69 seconds/step, 4.73 examples/second INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_162.jpg INFO:root:Step 011940, model loss 0.2865, total loss 0.5615, 1.70 seconds/step, 4.71 examples/second INFO:root:before shrink poly area:34.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7293.jpg INFO:root:Step 011950, model loss 0.7341, total loss 1.0091, 1.71 seconds/step, 4.68 examples/second INFO:root:Step 011960, model loss 0.7485, total loss 1.0235, 1.70 seconds/step, 4.71 examples/second INFO:root:Step 011970, model loss 0.3049, total loss 0.5799, 1.29 seconds/step, 6.22 examples/second INFO:root:Step 011980, model loss 0.1504, total loss 0.4255, 1.57 seconds/step, 5.11 examples/second INFO:root:Step 011990, model loss 0.2094, total loss 0.4844, 1.25 seconds/step, 6.39 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1320.jpg INFO:root:before shrink poly area:40.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1320.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1320.jpg INFO:root:before shrink poly area:30.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_427.jpg INFO:root:Step 012000, model loss 0.4236, total loss 0.6986, 1.50 seconds/step, 5.33 examples/second INFO:root:Step 012010, model loss 0.3278, total loss 0.6028, 1.49 seconds/step, 5.36 examples/second INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6840.jpg INFO:root:Step 012020, model loss 0.3612, total loss 0.6362, 1.46 seconds/step, 5.46 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7155.jpg INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6512.jpg INFO:root:Step 012030, model loss 0.1253, total loss 0.4003, 1.19 seconds/step, 6.73 examples/second INFO:root:before shrink poly area:26.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7643.jpg INFO:root:before shrink poly area:42.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_387.jpg INFO:root:before shrink poly area:78.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_387.jpg INFO:root:before shrink poly area:49.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_387.jpg INFO:root:Step 012040, model loss 0.5479, total loss 0.8229, 1.69 seconds/step, 4.74 examples/second INFO:root:Step 012050, model loss 0.5014, total loss 0.7764, 1.70 seconds/step, 4.71 examples/second INFO:root:Step 012060, model loss 0.9840, total loss 1.2590, 1.70 seconds/step, 4.71 examples/second INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:18.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_883.jpg INFO:root:Step 012070, model loss 0.2245, total loss 0.4995, 2.02 seconds/step, 3.95 examples/second INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1695.jpg INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1695.jpg INFO:root:Step 012080, model loss 0.1941, total loss 0.4691, 2.31 seconds/step, 3.46 examples/second INFO:root:before shrink poly area:42.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:22.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:22.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:before shrink poly area:52.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1564.jpg INFO:root:Step 012090, model loss 0.2164, total loss 0.4914, 2.40 seconds/step, 3.33 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1695.jpg INFO:root:Step 012100, model loss 0.3670, total loss 0.6420, 2.42 seconds/step, 3.30 examples/second INFO:root:Step 012110, model loss 0.2217, total loss 0.4967, 2.66 seconds/step, 3.01 examples/second INFO:root:Step 012120, model loss 0.1813, total loss 0.4562, 2.43 seconds/step, 3.30 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7876.jpg INFO:root:Step 012130, model loss 0.6228, total loss 0.8978, 2.43 seconds/step, 3.30 examples/second INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_677.jpg INFO:root:before shrink poly area:16.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part1/image_677.jpg INFO:root:Step 012140, model loss 0.2579, total loss 0.5329, 2.43 seconds/step, 3.29 examples/second INFO:root:before shrink poly area:39.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7702.jpg INFO:root:Step 012150, model loss 0.1355, total loss 0.4104, 2.52 seconds/step, 3.18 examples/second INFO:root:Step 012160, model loss 0.3326, total loss 0.6076, 2.57 seconds/step, 3.12 examples/second INFO:root:Step 012170, model loss 0.9681, total loss 1.2430, 2.57 seconds/step, 3.11 examples/second INFO:root:Step 012180, model loss 0.2479, total loss 0.5229, 2.63 seconds/step, 3.04 examples/second INFO:root:Step 012190, model loss 0.3430, total loss 0.6180, 2.67 seconds/step, 3.00 examples/second INFO:root:Step 012200, model loss 0.5021, total loss 0.7771, 2.66 seconds/step, 3.01 examples/second INFO:root:Step 012210, model loss 0.6480, total loss 0.9229, 3.02 seconds/step, 2.65 examples/second INFO:root:Step 012220, model loss 0.6268, total loss 0.9017, 2.69 seconds/step, 2.98 examples/second INFO:root:Step 012230, model loss 0.1907, total loss 0.4656, 2.68 seconds/step, 2.99 examples/second INFO:root:Step 012240, model loss 0.4899, total loss 0.7649, 2.70 seconds/step, 2.96 examples/second INFO:root:Step 012250, model loss 0.4158, total loss 0.6908, 2.66 seconds/step, 3.01 examples/second INFO:root:Step 012260, model loss 0.2279, total loss 0.5028, 2.64 seconds/step, 3.03 examples/second INFO:root:Step 012270, model loss 0.5709, total loss 0.8458, 2.74 seconds/step, 2.92 examples/second INFO:root:Step 012280, model loss 0.2614, total loss 0.5363, 2.68 seconds/step, 2.99 examples/second INFO:root:Step 012290, model loss 0.2808, total loss 0.5557, 2.73 seconds/step, 2.93 examples/second INFO:root:before shrink poly area:12.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part4/image_3497.jpg INFO:root:Step 012300, model loss 0.2940, total loss 0.5690, 2.68 seconds/step, 2.98 examples/second INFO:root:Step 012310, model loss 0.3482, total loss 0.6232, 2.95 seconds/step, 2.71 examples/second INFO:root:before shrink poly area:160.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7010.jpg INFO:root:before shrink poly area:174.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_7010.jpg INFO:root:Step 012320, model loss 0.2461, total loss 0.5210, 2.74 seconds/step, 2.92 examples/second INFO:root:Step 012330, model loss 0.5324, total loss 0.8074, 2.72 seconds/step, 2.95 examples/second INFO:root:Step 012340, model loss 0.2147, total loss 0.4896, 2.70 seconds/step, 2.96 examples/second INFO:root:Step 012350, model loss 0.1569, total loss 0.4318, 2.66 seconds/step, 3.00 examples/second INFO:root:Step 012360, model loss 0.2335, total loss 0.5084, 2.70 seconds/step, 2.96 examples/second INFO:root:before shrink poly area:10.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1495.jpg INFO:root:Step 012370, model loss 0.3350, total loss 0.6100, 2.73 seconds/step, 2.93 examples/second INFO:root:Step 012380, model loss 0.4588, total loss 0.7338, 2.69 seconds/step, 2.97 examples/second INFO:root:before shrink poly area:52.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_5191.jpg INFO:root:Step 012390, model loss 0.6016, total loss 0.8765, 2.65 seconds/step, 3.02 examples/second INFO:root:Step 012400, model loss 0.3943, total loss 0.6692, 2.77 seconds/step, 2.89 examples/second INFO:root:Step 012410, model loss 0.2016, total loss 0.4765, 3.00 seconds/step, 2.66 examples/second INFO:root:Step 012420, model loss 0.2786, total loss 0.5535, 2.75 seconds/step, 2.91 examples/second INFO:root:Step 012430, model loss 0.3956, total loss 0.6706, 2.73 seconds/step, 2.93 examples/second INFO:root:Step 012440, model loss 0.1406, total loss 0.4155, 2.77 seconds/step, 2.88 examples/second INFO:root:Step 012450, model loss 0.2955, total loss 0.5704, 2.72 seconds/step, 2.94 examples/second INFO:root:Step 012460, model loss 0.3479, total loss 0.6228, 2.74 seconds/step, 2.92 examples/second INFO:root:before shrink poly area:64.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part4/image_3043.jpg INFO:root:Step 012470, model loss 0.3258, total loss 0.6007, 2.67 seconds/step, 3.00 examples/second INFO:root:Step 012480, model loss 0.1647, total loss 0.4396, 2.76 seconds/step, 2.90 examples/second INFO:root:Step 012490, model loss 0.7657, total loss 1.0406, 2.77 seconds/step, 2.89 examples/second INFO:root:before shrink poly area:42.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1229.jpg INFO:root:before shrink poly area:32.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1229.jpg INFO:root:before shrink poly area:34.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_5252.jpg INFO:root:Step 012500, model loss 0.5924, total loss 0.8673, 2.78 seconds/step, 2.88 examples/second INFO:root:before shrink poly area:14.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part6/image_6627.jpg INFO:root:before shrink poly area:46.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part2/image_1364.jpg INFO:root:Step 012510, model loss 0.2573, total loss 0.5322, 3.06 seconds/step, 2.61 examples/second INFO:root:Step 012520, model loss 0.6056, total loss 0.8805, 2.68 seconds/step, 2.99 examples/second INFO:root:before shrink poly area:60.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_5358.jpg INFO:root:Step 012530, model loss 0.2255, total loss 0.5004, 2.76 seconds/step, 2.90 examples/second INFO:root:Step 012540, model loss 0.4113, total loss 0.6862, 2.69 seconds/step, 2.98 examples/second INFO:root:Step 012550, model loss 0.3312, total loss 0.6061, 2.69 seconds/step, 2.97 examples/second INFO:root:Step 012560, model loss 0.4030, total loss 0.6779, 2.73 seconds/step, 2.94 examples/second INFO:root:before shrink poly area:60.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part4/image_3246.jpg INFO:root:Step 012570, model loss 0.4791, total loss 0.7540, 2.71 seconds/step, 2.95 examples/second INFO:root:Step 012580, model loss 0.3284, total loss 0.6033, 2.81 seconds/step, 2.84 examples/second INFO:root:before shrink poly area:24.0 len(shrinked_poly) is 0,image ./icdar2017rctw_train_v1.2/train/part5/image_5191.jpg INFO:root:Step 012590, model loss 0.2097, total loss 0.4846, 2.69 seconds/step, 2.97 examples/second INFO:root:Step 012600, model loss 0.1623, total loss 0.4371, 2.72 seconds/step, 2.95 examples/second INFO:root:Step 012610, model loss 0.2951, total loss 0.5700, 2.98 seconds/step, 2.69 examples/second INFO:root:Step 012620, model loss 0.4073, total loss 0.6822, 2.80 seconds/step, 2.86 examples/second INFO:root:Step 012630, model loss 0.1932, total loss 0.4681, 2.70 seconds/step, 2.96 examples/second INFO:root:Step 012640, model loss 0.7217, total loss 0.9966, 2.73 seconds/step, 2.93 examples/second INFO:root:Step 012650, model loss 0.1983, total loss 0.4732, 2.80 seconds/step, 2.86 examples/second INFO:root:Step 012660, model loss 0.3414, total loss 0.6162, 2.71 seconds/step, 2.95 examples/second

XiaoLaoDi commented 5 years ago

So, what's wrong, how to fix that ?

XiaoLaoDi commented 5 years ago

And the text detect results of the newest trained model is so bad when compared to your pretrained model, which seems that the training does not really works.

XiaoLaoDi commented 5 years ago

And I am not sure whether the icdar2017 downloaded is correct, would you mind to share your training icdar2015 dataset or icdar2017 dataset ?

liuheng92 commented 5 years ago

I downloaded from icdar official website

lixingwei1106 commented 5 years ago

你好，我使用自己制作的数据集训练时，代码可以正常训练，但是model loss这一项一直为1，想请教一下你是如何训练自己的数据集的。

ganliqiang commented 5 years ago

And I am not sure whether the icdar2017 downloaded is correct, would you mind to share your training icdar2015 dataset or icdar2017 dataset ?

i meet the same error , do you fix them now?can you share the solution?

XiaoLaoDi commented 5 years ago

@lixingwei1106 @ganliqiang 我改用他官方的代码了，官网的实现没问题，也可以正常训练，只不过框架用的是PyTorch，链接是 https://github.com/whai362/PSENet

dsandii commented 5 years ago

@XiaoLaoDi @ganliqiang Did either of you resolve the issue of model not converging during training? I have been trying to understand what's wrong, but no success. Did you have to just try a different text detection implementation?

pingzi5233 commented 4 years ago

你好，我使用自己制作的数据集训练时，代码可以正常训练，但是model loss这一项一直为1，想请教一下你是如何训练自己的数据集的。

@lixingwei1106你好，我现在也遇到同样的问题，训练model loss一直为1，测试训练出来的模型结果与预识别前的一模一样，请问最后解决了吗？能否告知一二？