need help on dataset and inference code
Opened this issue · 2 comments
Hi, @jeff-j-chen , thanks for your amazing contribution on this project.
I am interested on the same.
can you provide me with more about the dataset collection(if possible provide some sample of dataset).
Anther query regrading the inference code:
Can you provide some snippet or any pipeline code that you used for inference using your trained .h5 model file.
Hope to hear from you soon.
Thanks and regards.
Hello, thanks for your interest with this project. This was made when I was new to machine learning, and I don't recommend you follow my steps. I would recommend looking at this (slightly) better repository, here.
However, if you still want to go for this: It's been a long time, so I'll try to do my best. From what I remember, I had code to process an entire video of someone sitting exclusively with a given posture (either good or bad). The camera angle and position was constantly changed throughout the video to constantly. During processing, the 10 keypoints are captured (visible in the video), and the points are collated into a single .npy file (you can see these in trainer.ipynb; each .npy file is the result of processing a video). As for the .h5 model file, that is generated by the trainer.ipynb.
Again, I highly recommend looking at my newer repository, the code is better (and commented)!
Thanks @jeff-j-chen , for your quick and prompt response.
Actually, I am looking for Sitting pose correcting solution.
Your work here is about whenever your whole body exposed to the camera.
But, when a portion of the body is exposed like while sitting in front the dash cam, that is of my primary concern.
Anyway, thanks for your quick and prompt help.
Looking forward to get some data output from you.