PackNet Training Based on our own Dataset and Question about train_kitti.yaml

TRI-ML / packnet-sfm

TRI-ML Monocular Depth Estimation Repository

https://tri-ml.github.io/packnet-sfm/

MIT License

1.24k stars 243 forks source link

PackNet Training Based on our own Dataset and Question about train_kitti.yaml #49

Closed soheilAppear closed 4 years ago

soheilAppear commented 4 years ago

Hi,

I wanted to use training part for my own dataset. However, I'm not sure what are the information do I need for the training part? Do we need point clouds in addition to the images of our own dataset?

also in the config files in train_kitti.yaml file:

" params: crop: 'garg' min_depth: 0.0 max_depth: 80.0 "

What are params and how did you determine the min and max depth? Is it only based on the Kitti dataset or something else?

Thank you so much! Soheil.

soheilAppear commented 4 years ago

e.g. is it possible also to train the model based on Tiny Kitti datasets also? If yes what part of the information is needed from the raw_kitti dataset?

soheilAppear commented 4 years ago

Is there any specific preparation that we should do on our own dataset in order to run the PackNet Training on our own dataset?

VitorGuizilini-TRI commented 4 years ago

Thank you for the interest in our work!

For self-supervised learning you only need to provide images, with the appropriate context (backward and forward), plus the camera intrinsics.
These parameters are for evaluation only, they determine which depth range and image area you consider when calculating the metrics.
We provide KITTI_tiny directly, you can download it and run any model, it's a quick way to see if your networks and losses are working (an overfit test).

porwalnaman01 commented 3 years ago

Hi, is there any way to predict camera intrinsics for every image in the dataset.?