Real-time depth extraction from image and video using with intel-isl/MiDaS depth extraction model. If you have CUDA support, I strongly recommend to use with GPU rather CPU.
Please refer to transformers.resize() for speed up the project.
dept_extraction_2.mp4
python midas_depth_extraction.py --video --path csgo_walk_2.mp4| Name | Type | Description |
|---|---|---|
| --video | bool | video frame depth extraction |
| --image | bool | image depth extraction |
| --path | str | Absolute path to the image/video |