https://github.com/clementsan/ir-tp-net
AI-based 3D depth sensing from stereo vision systems
https://github.com/clementsan/ir-tp-net
deep-learning depth-sensing pytorch stereo-vision torchio
Last synced: 2 months ago
JSON representation
AI-based 3D depth sensing from stereo vision systems
- Host: GitHub
- URL: https://github.com/clementsan/ir-tp-net
- Owner: clementsan
- Created: 2023-12-20T18:26:15.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-12-28T18:15:34.000Z (over 1 year ago)
- Last Synced: 2025-02-04T13:03:32.410Z (4 months ago)
- Topics: deep-learning, depth-sensing, pytorch, stereo-vision, torchio
- Language: Python
- Homepage:
- Size: 158 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# AI-based 3D depth sensing from stereo vision systems
Multi-input neural network to predict disparity maps
Network name: ir-tp-net (infrared tile processing network)
## Notes
### ir-tp-net
ir-tp-net is a two-stage deep neural network (DNN). Stage 1 includes dynamic parallel sub-networks, that are then concatenated to form a single sub-network in stage 2.
This python library works on specific multi-layer TIFF images, that were generated via a complex preprocessing
stage for multi-sensor stereo-vision (including image correction and 2D phase correlation).### Related github repo
The additional github repository ["disp-map-analysis"](https://github.com/clementsan/disp-map-analysis) provides additional pre-processing and post-processing python scripts.
## Installation and requirements
Installation using conda
```
> conda create --name --file
> conda activate
```
OR
```
> source activate
```General libraries being used:
- python 3.9
- pillow, numpy, pandas, matplotlib
- pytorch, torchio, imageio, tensorboard
- scikit-learn, yaml## Execution
### DNN training using configuration file
Command line:
```
> python3 AI_Training.py --config AI_Training_Config.yaml
```
### DNN inference using configuration fileCommand line:
```
> python3 AI_Inference_CSV.py --config AI_Inference_Config.yaml --verbose
```