Open lengyuner opened 3 years ago
Hi @lengyuner I would start by changing the dataset generator. It should return heatmaps and pafs adjusted to your number of channels (20) You also need to change the number of output channels in the model. The estimation part is the most complex, so you would have to make a lot of small adjustments here and there.
I will write a tutorial on this topic on my blog www.worldestimators.com
Thx~ I have changed the model successfully for my data and I may arrange the project pull it in my github.
I want to use this model for images of 20+ key points. What should I do?