megvii-research/OccDepth
Maybe the first academic open work on stereo 3D SSC method with vision-only input.
PythonApache-2.0
Issues
- 1
数据集训练问题
#20 opened by sorrymaker-h - 2
still onnx file export issue
#21 opened by tristanqiu8 - 0
Some problem about onnx
#19 opened by chi0612 - 0
Updated to CUDA11.8! Adapted to 4090!
#18 opened by HUANGYming - 0
Dimension Mismatch Error in models/unet2d.py
#17 opened by HUANGYming - 0
GPU memory requirements
#16 opened by Sxiansheng24 - 0
关于数据集的问题
#15 opened by zhengnx0906 - 0
Pre-trained model weights
#14 opened by nautilus-a - 0
预训练模型加载问题
#13 opened by Sxiansheng24 - 0
visiallization problem
#12 opened by Steven0928 - 2
Training result is terrible
#11 opened by Steven0928 - 2
Validating problem
#10 opened by Steven0928 - 1
Training is very slow.
#9 opened by If-only1 - 8
About Training processing
#3 opened by Louis-Leee - 0
- 6
- 5
Can't install mmcv properly.
#4 opened by JohnFuguiWang - 2
trained models(NYU_multicam_flospdepth_crp_stereodepth_cascadecls_2080ti_mIoU29.0.ckpt) is not match with model struct
#6 opened by xyzhou-dev - 0
无法导入MultiScaleDeformableAttention
#5 opened by Jessy-Huang - 3
Key 'data_root' is not in struct
#2 opened by Jessy-Huang