Face Detection With Eyes Movement Tracking
Eyes Movement Tracking
We have used multiple deep-learning models for eye movement tracking.
Firstly, the face is detected from the input video using a face detection model. The detected face is passed to the eye detection model and head pose estimation model. The head-pose angle and the detected left and right eyes are passed to the gaze estimation model to get the gaze vector.
One of the applications of gaze estimation is moving a computer pointer on the screen.
Eyes movement tracking output
Benchmarks on CPU
Include the benchmark results of running multiple model precisions. The Performance tests were run on Intel® Core™ i3-8350K CPU @ 4.00GHz × 4 and 16 GB Ram
| Properties | FP32 | FP16 | INT8 |
|---|---|---|---|
| Model Loading | 0.19139 | 0.183228 | 0.35699 |
| Infer Time | 75.9599 | 75.77568 | 68.5793 |
| FPS | 0.77672 | 0.77861fps | 0.8603fps |
Our customers are our biggest fans
Hear from the clients we have worked with
"I highly recommend LvisionAI for their efficient, motivated, and professional approach in deploying our model code into a functional model server, making our collaboration a positive experience.“"
Heraeus Consulting & IT Solutions GmbH
Germany
"I was interested in the practical implementation of deep learning plus computer vision and LvisionAI has helped me a great deal on this. Shahid and the team are very professional and accommodating to my needs. The communication was clear and all the deliverables are on time. I look forward to using LvisionAI service again and would recommend them to friends!"
Professor at University of Southampton
United Kingdom
"The LvisionAI team is skilled and reliable in developing custom deep learning applications. They are consistently hardworking in developing long term client projects. I would highly recommend them."
Canada