Failed - requirements installation

[Rut-B]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 5.4/5.4 MB 30.0 MB/s eta 0:00:00 hi @saidwivedi thanks of your repo. I trying to run your code and during running requirements installation get error of:

Installing build dependencies ... done Getting requirements to build wheel ... done error: subprocess-exited-with-error

× Preparing metadata (pyproject.toml) did not run successfully. │ exit code: 1 ╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip. Preparing metadata (pyproject.toml) ... error error: metadata-generation-failed

× Encountered error while generating package metadata. ╰─> See above for output.

note: This is an issue with the package mentioned above, not pip. hint: See above for details.

CalledProcessError Traceback (most recent call last) in <cell line: 1>()

[saidwivedi]

Can you share the entire log?

[Rut-B]

This is my colab notebook https://colab.research.google.com/drive/11Mk1VItszbAIXQGOoVreqZc7zrNzk94K?usp=sharing The last failing is

raceback (most recent call last): File "/content/POCO/demo.py", line 313, in main(args) File "/content/POCO/demo.py", line 167, in main detections = tester.run_detector(input_image_folder) File "/content/POCO/pocolib/core/tester.py", line 145, in run_detector mot = MPT( File "/usr/local/lib/python3.10/dist-packages/multi_person_tracker/mpt.py", line 52, in init self.detector = YOLOv3( TypeError: YoloV3() got an unexpected keyword argument 'device'

[Rut-B]

I trying to run the model without using yolov3 using maskrcnn seems that your model is not accurate

see example of image of shapy https://github.com/muelea/shapy/blob/master/samples/images/img_01.jpg with your code its return :

what I missed?

[saidwivedi]

Installation Failure: I checked the google colab you shared, I think the issue is due installation problem of YOLOv3. Can you install as per the instructions mentioned? Try this pip install git+https://github.com/mkocabas/yolov3-pytorch.git

Shape Estimation Like most methods in the HPS community, POCO focuses on pose estimation, largely due to the limited training data with ground-truth shape. Hence, output is mean shape. However, recent synthetic data like BEDLAM tries to address data scarcity. One can potentially use these data to model both pose and shape uncertainty.

Hope it answers your question.

[Rut-B]

@saidwivedi thanks of your answer - regard the shape estimation - is it the answer to model accuracy ? this image is the static position and should be simple example of person, seems that body measurements of is not correct is it cause that model not training on example of this body?

[saidwivedi]

Yes you are right, POCO is not trained on diverse body shapes as most of the data has mean shape. Hence, the shape is not accurate.