一個問題 坑了兩天 解決不了
訓練數據：1280*1856
# Train the head branches
# Passing layers="heads" freezes all layers except the head
# layers. You can also pass a regular expression to select
# which layers to train by name pattern.
model.train(dataset_train, dataset_val,
learning_rate=config.LEARNING_RATE,
epochs=1,
layers='heads')
image=skimage.color.gray2rgb(image)
Starting at epoch 0. LR=0.001
Checkpoint Path: E:\Mask_RCNN-master-try\logs\shapes20181118T2016\mask_rcnn_shapes_{epoch:04d}.h5
Selecting layers to train
fpn_c5p5 (Conv2D)
fpn_c4p4 (Conv2D)
fpn_c3p3 (Conv2D)
fpn_c2p2 (Conv2D)
fpn_p5 (Conv2D)
fpn_p2 (Conv2D)
fpn_p3 (Conv2D)
fpn_p4 (Conv2D)
In model: rpn_model
rpn_conv_shared (Conv2D)
rpn_class_raw (Conv2D)
rpn_bbox_pred (Conv2D)
mrcnn_mask_conv1 (TimeDistributed)
mrcnn_mask_bn1 (TimeDistributed)
mrcnn_mask_conv2 (TimeDistributed)
mrcnn_mask_bn2 (TimeDistributed)
mrcnn_class_conv1 (TimeDistributed)
mrcnn_class_bn1 (TimeDistributed)
mrcnn_mask_conv3 (TimeDistributed)
mrcnn_mask_bn3 (TimeDistributed)
mrcnn_class_conv2 (TimeDistributed)
mrcnn_class_bn2 (TimeDistributed)
mrcnn_mask_conv4 (TimeDistributed)
mrcnn_mask_bn4 (TimeDistributed)
mrcnn_bbox_fc (TimeDistributed)
mrcnn_mask_deconv (TimeDistributed)
mrcnn_class_logits (TimeDistributed)
mrcnn_mask (TimeDistributed)
C:\Users\Administrator\AppData\Roaming\Python\Python35\site-packages\tensorflow\python\ops\gradients_impl.py:108: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.
"Converting sparse IndexedSlices to a dense Tensor of unknown shape. "
Epoch 1/1
image_id 14
ERROR:root:Error processing image {'mask_path': 'train_data/cv2_mask/7.png', 'height': 1857, 'width': 1280, 'yaml_path': 'train_data/labelme_json/7_json/info.yaml', 'path': 'train_data/pic/7.jpg', 'source': 'shapes', 'id': 14}
Traceback (most recent call last):
File "E:\Mask_RCNN-master-try\mrcnn\model.py", line 1710, in data_generator
use_mini_mask=config.USE_MINI_MASK)
File "E:\Mask_RCNN-master-try\mrcnn\model.py", line 1213, in load_image_gt
mask, class_ids = dataset.load_mask(image_id)
File "<ipython-input-3-e3b8787a20d8>", line 80, in load_mask
labels = self.from_yaml_get_class(image_id)
File "<ipython-input-3-e3b8787a20d8>", line 13, in from_yaml_get_class
del labels[0]
KeyError: 0
---------------------------------------------------------------------------
KeyError Traceback (most recent call last)
<ipython-input-8-b3af8777b8e4> in <module>()
6 learning_rate=config.LEARNING_RATE,
7 epochs=1,
----> 8 layers='heads')
9
10 image=skimage.color.gray2rgb(image)
E:\Mask_RCNN-master-try\mrcnn\model.py in train(self, train_dataset, val_dataset, learning_rate, epochs, layers, augmentation, custom_callbacks, no_augmentation_sources)
2373 max_queue_size=100,
2374 workers=workers,
-> 2375 use_multiprocessing=True,
2376 )
2377 self.epoch = max(self.epoch, epochs)
C:\Users\Administrator\AppData\Roaming\Python\Python35\site-packages\keras\legacy\interfaces.py in wrapper(*args, **kwargs)
89 warnings.warn('Update your `' + object_name +
90 '` call to the Keras 2 API: ' + signature, stacklevel=2)
---> 91 return func(*args, **kwargs)
92 wrapper._original_function = func
93 return wrapper
C:\Users\Administrator\AppData\Roaming\Python\Python35\site-packages\keras\engine\training.py in fit_generator(self, generator, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch)
2143 batch_index = 0
2144 while steps_done < steps_per_epoch:
-> 2145 generator_output = next(output_generator)
2146
2147 if not hasattr(generator_output, '__len__'):
E:\Mask_RCNN-master-try\mrcnn\model.py in data_generator(dataset, config, shuffle, augment, augmentation, random_rois, batch_size, detection_targets, no_augmentation_sources)
1708 load_image_gt(dataset, config, image_id, augment=augment,
1709 augmentation=augmentation,
-> 1710 use_mini_mask=config.USE_MINI_MASK)
1711
1712 # Skip images that have no instances. This can happen in cases
E:\Mask_RCNN-master-try\mrcnn\model.py in load_image_gt(dataset, config, image_id, augment, augmentation, use_mini_mask)
1211 # Load image and mask
1212 image = dataset.load_image(image_id)
-> 1213 mask, class_ids = dataset.load_mask(image_id)
1214 original_shape = image.shape
1215 image, window, scale, padding, crop = utils.resize_image(
<ipython-input-3-e3b8787a20d8> in load_mask(self, image_id)
78 occlusion = np.logical_and(occlusion, np.logical_not(mask[:, :, i]))
79 labels = []
---> 80 labels = self.from_yaml_get_class(image_id)
81 labels_form = []
82 for i in range(len(labels)):
<ipython-input-3-e3b8787a20d8> in from_yaml_get_class(self, image_id)
11 temp = yaml.load(f.read())
12 labels = temp['label_names']
---> 13 del labels[0]
14 return labels
15
KeyError: 0
救命啊！
找到的原因1：json的y值大於720https://github.com/matterport/Mask_RCNN/issues/636 【IndexError: index 720 is out of bounds for axis 0 with size 720】
縮小圖像 重新標註
{
"flags": {},
"shapes": [
{
"label": "bread",
"line_color": null,
"fill_color": null,
"points": [
[
130,
0
],
]
}
],
"lineColor": [
0,
255,
0,
128
],
"fillColor": [
255,
0,
0,
128
],
"imagePath": "0.jpg",
"imageData":
沒用！
原因2：要標註成多邊形 重新標註 沒用！https://github.com/matterport/Mask_RCNN/issues/608 【KeyError: 'all_points_y'】
解決：
經過兩位大佬提醒，是標籤出了問題，對比可以運行的\train_data\labelme_json\0_json\info.yaml數據：
我：
跑通：
證明之前用的json to dataset不合適，導致標籤讀入錯誤，刪除之前的labelme，安裝labelme3.2
批量json to dataset：for /r 路徑 %i in (*.json) do labelme_json_to_dataset %i【參考自：https://blog.csdn.net/qq_15969343/article/details/79840517】
感謝大佬！感謝qq網友！感謝實驗室小夥伴！