<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/a2036146-29d6-4623-aba3-d151c7842ab9/Ellipse_37.png" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/a2036146-29d6-4623-aba3-d151c7842ab9/Ellipse_37.png" width="40px" /> Home

</aside>

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/3e5e17ea-2f99-4f21-b93e-5371dc5d73bd/Ellipse_37.png" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/3e5e17ea-2f99-4f21-b93e-5371dc5d73bd/Ellipse_37.png" width="40px" /> Publications

</aside>

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/b62fa618-419d-431d-b984-2f7aa459085e/Ellipse_37.png" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/b62fa618-419d-431d-b984-2f7aa459085e/Ellipse_37.png" width="40px" /> About

</aside>

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/ff9553a8-db2d-434b-96f6-573746cc6f41/Ellipse_37.png" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/ff9553a8-db2d-434b-96f6-573746cc6f41/Ellipse_37.png" width="40px" /> Now

</aside>

This project aims at reducing the latency of computation in video object detection. The technique takes advantage of the redundancy of information commonly present in image sequences. Also, the strategy mimics the behavior of the human vision, which dynamically ignores the object context when performing a focused tracking.

Also, a novel deep reinforcement learning technique has been developed, where an agent can learn not only based on a unique reward function, but on a distribution of them. This allows for conditioning the agent on a particular reward function on inference time.

The full paper can be found here.

The framework shows outstanding results on platforms with reduced parallelization capabilities, such as CPUs, achieving an average latency reduction up to 2.09×, and obtaining FPS rates similar to their equivalent GPU platform, at the cost of a 1.11× mAP reduction.

The code has been publicly released and a an open-source dataset has been provided.

https://vimeo.com/426725929

<aside> <img src="https://uploads-ssl.webflow.com/5e5fb69af39fe076e1ccecae/5eb5c2f9ddf6ad28af3d9cf7_logo.svg" alt="https://uploads-ssl.webflow.com/5e5fb69af39fe076e1ccecae/5eb5c2f9ddf6ad28af3d9cf7_logo.svg" width="40px" /> Twitter

</aside>

<aside> <img src="https://uploads-ssl.webflow.com/5e5fb69af39fe076e1ccecae/5eb5c2f917e59a4cf3c4ef37_linkedin.svg" alt="https://uploads-ssl.webflow.com/5e5fb69af39fe076e1ccecae/5eb5c2f917e59a4cf3c4ef37_linkedin.svg" width="40px" /> LinkedIn

</aside>

<aside> <img src="https://img2.freepng.es/20180615/jok/kisspng-bitbucket-version-control-atlassian-logo-bitbucket-server-5b238f65becf40.3268676715290571257816.jpg" alt="https://img2.freepng.es/20180615/jok/kisspng-bitbucket-version-control-atlassian-logo-bitbucket-server-5b238f65becf40.3268676715290571257816.jpg" width="40px" /> Bitbucket

</aside>

<aside> <img src="https://image.flaticon.com/icons/png/512/25/25231.png" alt="https://image.flaticon.com/icons/png/512/25/25231.png" width="40px" /> Github

</aside>

All rights reserved ©