Skip to content

Latest commit

 

History

History
35 lines (23 loc) · 948 Bytes

File metadata and controls

35 lines (23 loc) · 948 Bytes

G2VLM Reconstruction Evaluation

Our reconstruction evaluation code is based on the PI3 evaluation code. Please follow the instructions (link) to prepare the required data. To run the reconstruction evaluation for G2VLM, the following modifications are required:

  1. In mv_recon/eval.py,relpose/eval_angle.py,monodepth/infer.py,modify the path added to sys so that it points to the codebase root directory.

    import sys
    sys.path.append("/path/to/G2VLM")
  2. Specify the model checkpoint path under configs/model/default.yaml.

  3. To avoid import conflicts, save the required data under datas instead of data.

1. Monocular Depth Estimation

python monodepth/infer.py
python monodepth/eval.py

2. Relative Camera Pose Estimation (Angular Metrics)

python relpose/eval_angle.py

3. Multi-view Reconstruction (Point Map Estimation)

python mv_recon/eval.py