THU-MIG ORG

YOLOv10: Real-Time End-to-End Object Detection [NeurIPS 2024]

YOLOv10 is a real-time object detection model that draws labeled boxes around things it spots in images or video, running faster than earlier YOLO versions by skipping a post-processing cleanup step.

YOLOE: Real-Time Seeing Anything [ICCV 2025]

RepViT: Revisiting Mobile CNN From ViT Perspective [CVPR 2024] and RepViT-SAM: Towards Real-Time Segmenting Anything

LSNet: See Large, Focus Small [CVPR 2025]

针对pytorch模型的自动化模型结构分析和修改工具集，包含自动分析模型结构的模型压缩算法库

YOLO-UniOW: Efficient Universal Open-World Object Detection

The official implementation of our ECCV 2024 publication, PYRA (Parallel Yielding Re-Activation).

official repo for paper "[CLS] Token Tells Everything Needed for Training-free Efficient MLLMs"

PrefixKV: Adaptive Prefix KV Cache is What Vision Instruction-Following Models Need for Efficient Generation [NeurIPS 2025]

No description.

Official implementation for ICLR 2023 paper Consolidator: Mergeable Adapter with Grouped Connections for Visual Adaptation

Official implementation of "One-dimensional Adapter to Rule Them All: Concepts, Diffusion Models and Erasing Applications (CVPR2024 Highlight)"

Official Implementation of "Box-Level Active Detection (CVPR2023 Highlight)"

No description.