DeepPS2

DeepPS2: Revisiting Photometric Stereo using Two Differently Illuminated Images (Accepted in ECCV 2022)

Ashish Tiwari and Shanmuganathan Raman

This work addresses the PS2 problem (photometric Stereo with two images) under uncalibrated and self-supervised setting.

Overview

DeepPS2 is implemented in PyTorch and tested with Ubuntu (16.04).

Download the DiLiGenT Dataset and extract it into the data folder.

Use the following command to run the code:

python train.py --obj [object_name] --gpu_ids [gpu_id] --checkpoints_dir [path_to_save_chkpts] --save_dir [path_to_save_visual_results]

Note

This is the preliminary version of the code. More information coming soon!