We are a friendly computer vision interest group within MLC. We meet every Saturday 11 AM PDT to discuss papers from vision research. Occasionally we come together to collaborate on various projects or do workshops. We welcome anyone with an interest in this field so come ask questions, give feedback or participate in any other form.

Status: active
Discord Channel: #computer-vision
Calendar Link: click here!
Previous Session Recordings Link click here!
Formats Guest Speakers, Rapid Review (Review multiple papers), Role Playing, Conference Accepted Paper Rundown

Upcoming Session Information[edit]

Date 11/12/2022
Topic Plenoxels: Radiance Fields without Neural Networks
Format In-depth review

Previous Reading Group Sessions[edit]

Date 10/29/2022
Topic Where Should I Spend My FLOPS? Efficiency Evaluations of Visual Pre-training Methods
Format In-depth review

We've reviewed/discussed >50 papers. Some of the papers we have discussed in the past are:

  • EG3D: Efficient Geometry-aware 3D Generative Adversarial Networks
  • Barlow Twins: Self-Supervised Learning via Redundancy Reduction
  • Show, Attend and Tell
  • AdaMML: Adaptive Multi-Modal Learning for Efficient Video Recognition
  • Viewmaker Networks: Learning Views for Unsupervised Representation Learning
  • Neural Discrete Representation Learning(VQ-VAE)
  • Input-level Inductive Biases for 3D Reconstruction
  • Object-aware Contrastive Learning for Debiased Scene Representation
  • Visual Chirality
  • Exploring Simple Siamese Representation Learning
  • An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale, and many more!
  • Swin Transformer -- Hierarchical Vision Transformer using Shifted Windows

A record of each session can also be found in this spreadsheet (In progress since 7/12/2022) If you have a paper suggestion that you'd like to present or discuss, come and join us on discord :)

Past Projects[edit]

Title Description Deadline
ICLR Blog Post Select a previous ICLR paper and write an informal and accessible blog post. Highlight ambiguities, provide better visualization, etc. Jan 14th, 2022
ML Reproducibility Challenge 2021 Select a paper, reproduce the major results, and verify the contribution. Feb 4th, 2022

This article was last modified: Nov. 6, 2022, 3:47 p.m. UTC

Powered by django-wiki, an open source application under the GPLv3 license.