search

ArminMasoumian / GCNDepth

Self-Supervised CNN-GCN Autoencoder for Monocular Depth Estimation
MIT License
113 stars 17 forks source link

explain the API #21

Closed zsh1223 closed 1 year ago

zsh1223 commented 1 year ago

Hi, could you explain the inferenceinfer_singleimage.py，eval_depth_pp.py，draw_odometry.py mean anything?

image

ArminMasoumian commented 1 year ago

draw_odometry.py: Analytical curves and KITTI odometry results. eval_pose.py: KITTI odometry results. eval_depth.py: KITTI depth results. infer.py: Generate depth maps. infer_singleimage.py: Test a single image.

zsh1223 commented 1 year ago

Thank you for your explanation of the API. Is infer_singleimage.py use the posenet networks? If use it, how does the posenet work in infer_singleimage.py?

ArminMasoumian commented 1 year ago

Thank you for your explanation of the API. Is infer_singleimage.py use the posenet networks? If use it, how does the posenet work in infer_singleimage.py?

Yes, we do use PoseNet in our network, and it plays a crucial role in generating depth images from a single picture. To achieve this, we require the necessary transposed information. For a detailed explanation, please refer to our paper at this link: https://www.sciencedirect.com/science/article/abs/pii/S0925231222013601.