search

TannerGilbert / Tensorflow-Object-Detection-API-Train-Model

Train a object detection model with the Tensorflow Object Detection API and Tensorflow 2.
https://gilberttanner.com/blog/creating-your-own-objectdetector
MIT License
194 stars 104 forks source link

label map path issue #17

Closed RaymondMcCreesh closed 3 years ago

RaymondMcCreesh commented 3 years ago

Having changed all lines in the config file, when I run the training i get the following error which makes me think there is another label map path being called?

image

this is my input image

TannerGilbert commented 3 years ago

Are you sure you changed all the PATH_TO_BE_CONFIGURED occurrences inside the training&efficientdet_d0_coco17_tpu-32/pipeline.config file?

RaymondMcCreesh commented 3 years ago

I found it was pointing to the wrong pythonpath after i reopened a terminal. Unfortunately I found a new issue which I have not been able to fix the past two days.
context dataset = dataset_fn(ctx) File "C:\Users\Ray\FYP\models\research\object_detection\model_lib_v2.py", line 518, in train_dataset_fn input_context=input_context) File "C:\Users\Ray\FYP\models\research\object_detection\inputs.py", line 875, in train_input reduce_to_frame_fn=reduce_to_frame_fn) File "C:\Users\Ray\FYP\models\research\object_detection\builders\dataset_builder.py", line 195, in build decoder = decoder_builder.build(input_reader_config) File "C:\Users\Ray\FYP\models\research\object_detection\builders\decoder_builder.py", line 63, in build load_track_id=input_reader_config.load_track_id) File "C:\Users\Ray\FYP\models\research\object_detection\data_decoders\tf_example_decoder.py", line 393, in init default_value=''), File "C:\Users\Ray\FYP\models\research\object_detection\data_decoders\tf_example_decoder.py", line 89, in init label_map_proto_file, use_display_name=False) File "C:\Users\Ray\FYP\models\research\object_detection\utils\label_map_util.py", line 201, in get_label_map_dict label_map = load_labelmap(label_map_path_or_proto) File "C:\Users\Ray\FYP\models\research\object_detection\utils\label_map_util.py", line 168, in load_labelmap label_map_string = fid.read() File "C:\Users\Ray\anaconda3\envs\FYP\lib\site-packages\tensorflow\python\lib\io\file_io.py", line 117, in read self._preread_check() File "C:\Users\Ray\anaconda3\envs\FYP\lib\site-packages\tensorflow\python\lib\io\file_io.py", line 80, in _preread_check compat.path_to_str(self.__name), 1024 * 512) tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: training_demo/annotations/label_map.pbtxt : The system cannot find the file specified. ; No such file or directory

I am runnning Annaconda in Windows 10 and get this issue whether I insert the full path or not. I'm not sure if the issue is that it can't see the file or it can't open the file?

TannerGilbert commented 3 years ago

If the problem is that it doesn't have the rights to open the file, you could try opening your console as administrator, but I don't think this is the problem. Are you sure you specified the correct path?

RaymondMcCreesh commented 3 years ago

my solution to this was to change the labelmap extension to just .txt and the system found the file

TannerGilbert commented 3 years ago

Good to hear that it's working now.