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" provides additional pre-processing and post-processing python scripts.

Installation and requirements

Installation using conda

> conda create --name <env_name> --file <conda_environment.yaml>
> conda activate <env_name>

OR

> source activate <env_name>

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 file

Command line:

> python3 AI_Inference_CSV.py --config AI_Inference_Config.yaml --verbose