'It looks like you are subclassing Model and you '

[ray-lee-94]

I would like to retrain the maskrcnn from the released model. But I got errors below:

---

/home/wen/anaconda3/bin/python /home/wen/net_project/MASK/Mask_RCNN-master/samples/coco/coco.py train --dataset=/home/wen/net_project/MASK/data/ --model=/home/wen/net_project/MASK/Mask_RCNN-master/weight/resnet50_coco_v0.1.0.h5
Using TensorFlow backend.
Command: train
Model: /home/wen/net_project/MASK/Mask_RCNN-master/weight/resnet50_coco_v0.1.0.h5
Dataset: /home/wen/net_project/MASK/data/
Year: 2014
Logs: /home/wen/net_project/MASK/Mask_RCNN-master/logs
Auto Download: False

Configurations:
BACKBONE resnet50
BACKBONE_STRIDES [4, 8, 16, 32, 64]
BATCH_SIZE 8
BBOX_STD_DEV [0.1 0.1 0.2 0.2]
COMPUTE_BACKBONE_SHAPE None
DETECTION_MAX_INSTANCES 100
DETECTION_MIN_CONFIDENCE 0.7
DETECTION_NMS_THRESHOLD 0.3
FPN_CLASSIF_FC_LAYERS_SIZE 1024
GPU_COUNT 4
GRADIENT_CLIP_NORM 5.0
IMAGES_PER_GPU 2
IMAGE_MAX_DIM 1024
IMAGE_META_SIZE 93
IMAGE_MIN_DIM 800
IMAGE_MIN_SCALE 0
IMAGE_RESIZE_MODE square
IMAGE_SHAPE [1024 1024 3]
LEARNING_MOMENTUM 0.9
LEARNING_RATE 0.001
LOSS_WEIGHTS {'rpn_class_loss': 1.0, 'rpn_bbox_loss': 1.0, 'mrcnn_class_loss': 1.0, 'mrcnn_bbox_loss': 1.0, 'mrcnn_mask_loss': 1.0}
MASK_POOL_SIZE 14
MASK_SHAPE [28, 28]
MAX_GT_INSTANCES 100
MEAN_PIXEL [123.7 116.8 103.9]
MINI_MASK_SHAPE (56, 56)
NAME coco
NUM_CLASSES 81
POOL_SIZE 7
POST_NMS_ROIS_INFERENCE 1000
POST_NMS_ROIS_TRAINING 2000
ROI_POSITIVE_RATIO 0.33
RPN_ANCHOR_RATIOS [0.5, 1, 2]
RPN_ANCHOR_SCALES (32, 64, 128, 256, 512)
RPN_ANCHOR_STRIDE 1
RPN_BBOX_STD_DEV [0.1 0.1 0.2 0.2]
RPN_NMS_THRESHOLD 0.7
RPN_TRAIN_ANCHORS_PER_IMAGE 256
STEPS_PER_EPOCH 1000
TOP_DOWN_PYRAMID_SIZE 256
TRAIN_BN False
TRAIN_ROIS_PER_IMAGE 200
USE_MINI_MASK True
USE_RPN_ROIS True
VALIDATION_STEPS 50
WEIGHT_DECAY 0.0001

Traceback (most recent call last):
File "/home/wen/anaconda3/lib/python3.6/site-packages/keras/engine/network.py", line 313, in setattr
is_graph_network = self._is_graph_network
File "/home/wen/net_project/MASK/Mask_RCNN-master/mrcnn/parallel_model.py", line 46, in getattribute
return super(ParallelModel, self).getattribute(attrname)
AttributeError: 'ParallelModel' object has no attribute '_is_graph_network'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/home/wen/net_project/MASK/Mask_RCNN-master/samples/coco/coco.py", line 455, in
model_dir=args.logs)
File "/home/wen/net_project/MASK/Mask_RCNN-master/mrcnn/model.py", line 1848, in init
self.keras_model = self.build(mode=mode, config=config)
File "/home/wen/net_project/MASK/Mask_RCNN-master/mrcnn/model.py", line 2073, in build
model = ParallelModel(model, config.GPU_COUNT)
File "/home/wen/net_project/MASK/Mask_RCNN-master/mrcnn/parallel_model.py", line 35, in init
self.inner_model = keras_model
File "/home/wen/anaconda3/lib/python3.6/site-packages/keras/engine/network.py", line 316, in setattr
'It looks like you are subclassing Model and you '
RuntimeError: It looks like you are subclassing Model and you forgot to call super(YourClass, self).__init__(). Always start with this line.

Process finished with exit code 1

the tensorflow is 1.10. Keras is 2.2.2. when I degrade Keras to 2.1.3, another error occured.

Loading weights /home/wen/net_project/MASK/Mask_RCNN-master/weight/resnet50_coco_v0.1.0.h5
Traceback (most recent call last):
File "/home/wen/net_project/MASK/Mask_RCNN-master/samples/coco/coco.py", line 474, in
model.load_weights(model_path, by_name=True)
File "/home/wen/net_project/MASK/Mask_RCNN-master/mrcnn/model.py", line 2141, in load_weights
saving.load_weights_from_hdf5_group_by_name(f, layers)
File "/home/wen/anaconda3/lib/python3.6/site-packages/keras/engine/topology.py", line 3233, in load_weights_from_hdf5_group_by_name
' element(s).')
ValueError: Layer #2 (named "conv1") expects 2 weight(s), but the saved weights have 1 element(s).

---

Maybe that the pretrained model dosen't fit the network?

[ray-lee-94]

There is no reply. I switch to single GPU that the error disappear.

[hgaiser]

Hey, sorry about that. multi-gpu on keras-maskrcnn isn't supported at the moment. It's also not in the train.py actually, how did you enable it?