Error in loading weights

[cppntn]

raise ValueError(err.message)
ValueError: Dimension 1 in both shapes must be equal, but are 92 and 84 for 'Assign_682' (op: 'Assign') with input shapes: [1024,92], [1024,84].

Any help would be appreciated,
Thanks!

[hanskrupakar]

This error is usually related to the number of classes in the dataset. Can you give me some more specifics? What did you try to run?

[cppntn]

I ran with a class file .txt made only by three classes. What exactly is the txt classe file that i should use?
Thanks

[cppntn]

Hi @hanskrupakar

I ran the command :

python segment.py -i set/ -f -p ../Mask_RCNN/ -w pretrained_weights/coco_10/mask_rcnn_bags_0006.h5 -c classes/products.txt

But this still gives me the error about the weights:

Traceback (most recent call last):
File "segment.py", line 434, in
gen = COCO_dataset_generator(fig, ax, args)
File "segment.py", line 110, in init
model.load_weights(args['weights_path'], by_name=True)
File "../Mask_RCNN/model.py", line 2105, in load_weights
topology.load_weights_from_hdf5_group_by_name(f, layers)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/keras/engine/topology.py", line 3248, in load_weights_from_hdf5_group_by_name
K.batch_set_value(weight_value_tuples)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py", line 2365, in batch_set_value
assign_op = x.assign(assign_placeholder)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/ops/variables.py", line 573, in assign
return state_ops.assign(self._variable, value, use_locking=use_locking)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/ops/state_ops.py", line 276, in assign
validate_shape=validate_shape)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/ops/gen_state_ops.py", line 57, in assign
use_locking=use_locking, name=name)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
op_def=op_def)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 2958, in create_op
set_shapes_for_outputs(ret)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 2209, in set_shapes_for_outputs
shapes = shape_func(op)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 2159, in call_with_requiring
return call_cpp_shape_fn(op, require_shape_fn=True)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/common_shapes.py", line 627, in call_cpp_shape_fn
require_shape_fn)
File "/home/addfor/anaconda3/envs/coco/lib/python3.6/site-packages/tensorflow/python/framework/common_shapes.py", line 691, in _call_cpp_shape_fn_impl
raise ValueError(err.message)
ValueError: Dimension 1 in both shapes must be equal, but are 92 and 84 for 'Assign_682' (op: 'Assign') with input shapes: [1024,92], [1024,84].

[hanskrupakar]

Oh I think the problem is that you're trying to get initial segmentations for a custom dataset (with different number of classes) using the pre-trained model for the products dataset used in this repo. You can generate the initial segmentations using Mask RCNN model for your custom dataset only if you train a Mask RCNN model for it.