search

marcelampc / d3net_depth_estimation

Dense Deep Depth Estimation Network (D3-Net) in PyTorch.
115 stars 19 forks source link

the interval of inputs and ground truth #5

Closed anmeizhang closed 5 years ago

anmeizhang commented 5 years ago

My depth ground truth are true depth values (from 0m to 10m), but the outputs of tanh are within [-1, 1], I notice your dataset is also NYU v2. So I want to ask what is your depth ground truth? By the way, are your input images in the interval [0, 1] or [0, 255]? I am looking forward for your answers. Thank you!

marcelampc commented 5 years ago

Hi! You may just delete the Tanh layer! I was working with GAN algorithms at first and used the Tanh to have values between -1 and 1 for the discriminator. I stopped using this layer now to output value in meters and save them in millimeters.