piperod / beepose Goto Github PK

When running:
python beepose/inference/inference_video.py --video VolumeData/videos/C02_170622140000.mp4 --model VolumeData/OUTPUT2/Inference_model.h5 --model_config VolumeData/OUTPUT2/model_params.json --GPU 0 --gpu_fraction 0.9 --end 1000 --model VolumeData/OUTPUT2/model_params.json --output VolumeData/OUTPUT2/test2.json
get the following two errors
AttributeError: 'Namespace' object has no attribute 'sufix'
AttributeError: 'Namespace' object has no attribute 'sufix'

I tried using the model I have trained but I get the following error while running the process_folder_images.py -
"Dimension 0 in both shapes must be equal, but are 1 and 28. Shapes are [1,1,512,12] and [28,512,1,1]. for 'Assign_40' (op: 'Assign') with input shapes: [1,1,512,12], [28,512,1,1]."
Here is the stack trace
Traceback (most recent call last):
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1659, in _create_c_op
c_op = c_api.TF_FinishOperation(op_desc)
tensorflow.python.framework.errors_impl.InvalidArgumentError: Dimension 0 in both shapes must be equal, but are 1 and 28. Shapes are [1,1,512,12] and [28,512,1,1]. for 'Assign_40' (op: 'Assign') with input shapes: [1,1,512,12], [28,512,1,1].

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "/usr/lib/python3.6/runpy.py", line 193, in _run_module_as_main
"main", mod_spec)
File "/usr/lib/python3.6/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/home/mrwick/beepose/beepose/inference/process_folder_image.py", line 61, in
model.load_weights(keras_weights_file)
File "/home/mrwick/venv/lib/python3.6/site-packages/keras/engine/topology.py", line 2645, in load_weights
load_weights_from_hdf5_group(f, self.layers)
File "/home/mrwick/venv/lib/python3.6/site-packages/keras/engine/topology.py", line 3166, in load_weights_from_hdf5_group
K.batch_set_value(weight_value_tuples)
File "/home/mrwick/venv/lib/python3.6/site-packages/keras/backend/tensorflow_backend.py", line 2365, in batch_set_value
assign_op = x.assign(assign_placeholder)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/ops/variables.py", line 1762, in assign
name=name)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/ops/state_ops.py", line 223, in assign
validate_shape=validate_shape)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/ops/gen_state_ops.py", line 64, in assign
use_locking=use_locking, name=name)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/util/deprecation.py", line 507, in new_func
return func(*args, **kwargs)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 3300, in create_op
op_def=op_def)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1823, in init
control_input_ops)
File "/home/mrwick/venv/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1662, in _create_c_op
raise ValueError(str(e))
ValueError: Dimension 0 in both shapes must be equal, but are 1 and 28. Shapes are [1,1,512,12] and [28,512,1,1]. for 'Assign_40' (op: 'Assign') with input shapes: [1,1,512,12], [28,512,1,1].

Hi, I have been trying to train on 2 gpus but I dont know how to specify both the GPUs. In the help it says if "all" then try to allocate on every gpu but using 'all' gives an error : ValueError : invalid literal for int() with base 10: all. Also specifying two numbers doesn't work. Can you please tell me how to use multiple GPUs ?

During training, I have observed that there is a huge difference between the L1_acc and L2_acc of any stage. While L2_acc has gone over 0.99 and is steadily increasing, L1_acc is unstable as it increases and decreases without any pattern. As I understand, L1_acc is about the groundtruths and L2_acc is for PAFs. Is this behavior possible and is the accuracy of groundtruths not related to PAFs ?

I have three questions regarding the mask images -

1. Does the function get_masks(self, annotations) in pose_dataset.py produces masks for each images using all the annotations associated with that image at a time ?
2. what is the purpose of mask_miss ?
3. Does the function not store the masks like in your dataset but only use them while running ?

Warning message at inference does not specify the connection type.

What is the score array for in calculate_peaks function in inference.py ?
score=[0.2,0.2,0.2,0.2,0.2,0.5,0.5,0.5,0.5].
I thought it was a threshold for detection of parts. If so, how are the values decided for each part and why are there 9 values in the list when the number of parts by default is 5 ?