c parameter in reverse Huber loss

yuhsuanyeh / BiFuse

[CVPR2020] BiFuse: Monocular 360 Depth Estimation via Bi-Projection Fusion

MIT License

173 stars 28 forks source link

c parameter in reverse Huber loss #10

Open capstayn opened 3 years ago

capstayn commented 3 years ago

Hi, I am implementing the loss described in your paper, in equation (5) (reverse Huber Loss). Would it be possible to know what value are you using for the c parameter in equation (4)?

capstayn commented 3 years ago

Just for the record, in the cited paper Deeper Depth Prediction with Fully Convolutional Residual Networks I found c defined as 0.2*max(absolute_diff)