lemariva / MaixPy_YoloV2

YOLOv2 object detector training for a MAix-board
https://lemariva.com/blog/2020/01/maixpy-object-detector-mobilenet-and-yolov2-sipeed-maix-dock
Apache License 2.0
23 stars 7 forks source link

Train on Google Colab not working #4

Open PhilippTriet opened 4 years ago

PhilippTriet commented 4 years ago

I always get the following error message when I try to train it on Google Colab. It works local and on Azure.

RuntimeError Traceback (most recent call last)

in () 24 config['model']['class_scale'], 25 config['model']['object_scale'], ---> 26 config['model']['no_object_scale']) 27 28 # 2. Load the pretrained weights (if any) 5 frames /content/drive/My Drive/Colab Notebooks/yolo/frontend.py in create_yolo(architecture, labels, input_size, anchors, coord_scale, class_scale, object_scale, no_object_scale) 33 n_classes = len(labels) 34 n_boxes = int(len(anchors)/2) ---> 35 yolo_network = create_yolo_network(architecture, input_size, n_classes, n_boxes) 36 yolo_loss = YoloLoss(yolo_network.get_grid_size(), 37 n_classes, /content/drive/My Drive/Colab Notebooks/yolo/backend/network.py in create_yolo_network(architecture, input_size, nb_classes, nb_box) 17 input_size, 18 nb_classes, ---> 19 nb_box) 20 return yolo_net 21 /content/drive/My Drive/Colab Notebooks/yolo/backend/network.py in __init__(self, feature_extractor, input_size, nb_classes, nb_box) 38 grid_size = 7 39 ---> 40 grid_size = feature_extractor.get_output_size() 41 42 print(grid_size) /content/drive/My Drive/Colab Notebooks/yolo/backend/utils/feature.py in get_output_size(self) 59 60 def get_output_size(self): ---> 61 output_shape = self.feature_extractor.get_output_shape_at(-1) 62 assert output_shape[1] == output_shape[2] 63 return output_shape[1] /usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in get_output_shape_at(self, node_index) 2028 """ 2029 return self._get_node_attribute_at_index(node_index, 'output_shapes', -> 2030 'output shape') 2031 2032 @doc_controls.do_not_doc_inheritable /usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py in _get_node_attribute_at_index(self, node_index, attr, attr_name) 2601 if not self._inbound_nodes: 2602 raise RuntimeError('The layer has never been called ' -> 2603 'and thus has no defined ' + attr_name + '.') 2604 if not len(self._inbound_nodes) > node_index: 2605 raise ValueError('Asked to get ' + attr_name + ' at node ' + RuntimeError: The layer has never been called and thus has no defined output shape.