GithubHelp home page GithubHelp logo

dpt's Introduction

Vision Transformers for Dense Prediction

This repository contains code and models for the Master's Thesis "Estimación de Profundidad Online con Transformers Eficientes", which modifies the code published along the paper presented by Ranftl et al. to accelerate the inference speed of the Dense Prediction Transformers.

Related links

Abstract

Monocular depth estimation deals with the automatic recovery of an approximation of the dimension that is lost when projecting a three-dimensional scene into a two-dimensional image. This problem has an infinite number of geometric solutions, which makes it practically impossible to solve using traditional computer vision techniques. Nonetheless, Deep Learning techniques are capable of extracting different characteristics from the images that make it possible to approximate a solution. In this work this problem and the existing solutions are studied, especially those based on Transformers and supervised learning. In one of these solutions, a series of modifications and developments are carried out to reduce the size of the original model and multiply its inference speed by nearly five. Furthermore, an exhaustive study, both quantitative and qualitative, of the influence of the different modifications is included, evaluating the models in the KITTI dataset, oriented to autonomous driving.

Documentation

Documentation for this project can be found in the Appendix B of the Master's Thesis manuscript (ES).

Acknowledgements

This work obviously would not have been possible without the incredibly valuable contribution of the Vision Transformers for Dense Prediction paper and the implementations of efficient attention mechanisms from Phil Wang. Likewise, a huge thank you to the PyTorch community and Ross Wightman for his incredible work with timm.

License

MIT License

dpt's People

Contributors

alexeyab avatar barnjamin avatar dvdhfnr avatar guillesanbri avatar ranftlr avatar timmh avatar xiong-jie-y avatar

Watchers

 avatar

Forkers

simonsroad

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.