Link: https://cvg.cit.tum.de/teaching/ss2025/dl4sai

Organisers

Project examples

  • use of the SOTA model called DUSt3R
  • the model basically does the registration between two images.
  • the idea: use the model to do Simultanious Location and Mapping;

  • Use depth information to do semantic instance segmentation;

  • use a single view reconstruction network that reconstructs highlighted object

Prerequisites

  • They are rather soft, if you have background equivalent to taking listed courses is fine too;
  • If you are willing to get up to speed that would be also alright;
  • If unsure send an email;

Timeline

  • first cast you vote for courses via the matching system;
  • then the organisers will also cast they votes;
  • matching system does it’s magic;
  • after matching the projects are presented and you choose and pick;
  • during midterm talks you can decide would

You are graded on:

  • two presentations ;
  • and a written project;

Questions and aswers

  1. Send the CV/transcript to the dl4sai-ss25@vision.in.tum.de
  2. Q: Is it publishable? A: It’s not enough time to publish it. Usually, you would have to put some more time on top of the project; If you are interested in doing so, say it to the advisor. Such that they can align the project goals;
  3. Q: What was the acceptance rate last year? A: 30%
  4. Q: Can I send the motivation letter? A: Yes