search

wayveai / fiery

PyTorch code for the paper "FIERY: Future Instance Segmentation in Bird's-Eye view from Surround Monocular Cameras"
https://wayve.ai/blog/fiery-future-instance-prediction-birds-eye-view
MIT License
556 stars 85 forks source link

how did you project the estimated segmentation on the corresponding RGB images? #10

Open sunnyHelen opened 3 years ago

sunnyHelen commented 3 years ago

Hi, thanks for sharing your great work. I saw the predicted segmentation is projected on the RGB images in Figure 1 in your paper. Some are dotted and some are transparent squares. What's the meaning of them? And how to visualize them?

anthonyhu commented 3 years ago

Hey!

Thanks for the kind words :) The dots correspond to the bird's-eye view vehicle segmentation that has been discretised and projected back into the RGB image. The transparent paths correspond to the future trajectories in bird's-eye view projected to RGB. To obtain these, you need to use the intrinsics and extrinsics to go from bird's-eye view space to image space.

sunnyHelen commented 3 years ago

Thanks a lot for your quick reply. The visualizing code in this repo doesn't include this part right?

sunnyHelen commented 3 years ago

Could you please help me figure out how to go from bird's-eye view space to image space in detail?

sunnyHelen commented 3 years ago

I wrote this part by myself, but it didn't work. Could you please release this visualization part? This is very important to me.