Computer Vision

Detection, Pose Estimation and Segmentation for Multiple Bodies: Closing the Virtuous Circle
https://github.com/MiraPurkrabek/BBoxMaskPose

OneVision Encoder
https://github.com/EvolvingLMMs-Lab/OneVision-Encoder

Qwen-Image-Layered
https://github.com/QwenLM/Qwen-Image-Layered

In Pursuit of Pixel Supervision for Visual Pre-training
https://github.com/facebookresearch/pixio

AnyUp: Universal Feature Upsampling
https://github.com/wimmerth/anyup

FoundationStereo: Zero-Shot Stereo Matching
https://github.com/NVlabs/FoundationStereo/

Depth Anything 3
https://github.com/ByteDance-Seed/Depth-Anything-3

Qwen-Image-Lightning
https://github.com/ModelTC/Qwen-Image-Lightning/

EdgeTAM: On-Device Track Anything Model
https://github.com/facebookresearch/EdgeTAM

Image Matching Models
https://github.com/alexstoken/image-matching-models

Pixel-Perfect Depth
https://github.com/gangweix/pixel-perfect-depth

Into the Rabbit Hull: From Task-Relevant Concepts in DINO to Minkowski Geometry
https://arxiv.org/pdf/2510.08638

SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization
https://huggingface.co/facebook/SSDD

3D Human Motion Generation and Simulation Tutorial
https://3dmogen.github.io/

DINO-Foresight: Looking into the Future with DINO
https://github.com/Sta8is/DINO-Foresight

AToken: A Unified Tokenizer for Vision
https://arxiv.org/abs/2509.14476