Computer-Human Interface Peripheral Signal Manipulation via AI Retina Tracking (CHIPSMART)
Requires:
- One or both of the following hardware:
- Intel® CPU
- Intel® Neural Compute Stick 2
- All of the following software:
- Ubuntu 18.04.4 LTS (Bionic Beaver)
- Python 3.7.6
- Intel® Distribution of OpenVINO Toolkit 2020.3 (LTS)
- This repository, cloned to a local directory
To run this project, simply execute the script titled demo.sh
in the root directory of this repository:
./demo.sh
This readme file
The Intel® hardware I benchmarked with:
Model Loading Times (ms)
- Face Detection: 1117.04
- Facial Landmark Detection: 443.23
- Gaze Estimation: 326.16
- Head Pose Estimation: 145.82
- Total Loading Time (All Models): 1117.99
Model Inference Times (ms)
- Face Detection: 102.60
- Facial Landmark Detection: 1.55
- Gaze Estimation: 9.10
- Head Pose Estimation: 7.14