Center Score Map Prediction of Neural Network not Optimized

[kaixin-bai]

In my implementation of the neural network described in your paper, the network predicts a center score map as a part of its output. However, when using my own dataset and modifying the depth map range to match the range in the dataset, I noticed that the predicted center score map is not performing optimally.

I would appreciate your guidance on how to optimize the center score map prediction of the neural network. Specifically, what are some techniques or approaches that could improve the accuracy of the predicted center score map?

Thank you for your time and expertise.

The center score map:

the mix map:

sampled:

smoothness:

ess: