All-in-one training for vision models (YOLO, ViTs, RT-DETR, DINOv3): pretraining, fine-tuning, distillation.

SimpleAICV:pytorch training examples.

Pytorch implementation of "EdgeCrafter: Compact ViTs for Edge Dense Prediction via Task-Specialized Distillation"

A repository to apply DINOv3 models for different downstream tasks: image classification, semantic segmentation, object detection.

ROS 2 integration of Meta’s DINOv3 backbone with lightweight heads for vision tasks.

dinov3例子

Integrating SAM2 with DINOv2/v3 for segmentation

Gaze-LLE-DINOv3: Gaze Target Estimation via Large-Scale Learned Encoders with DINOv3.

Command-line tool for extracting DINOv3, CLIP, SigLIP2, RADIO, features for images and videos

[CVPR'26 Highlight] Official Code for “V²-SAM: Marrying SAM2 with Multi-Prompt Experts for Cross-View Object Correspondence”

Switch the backbone of mask2former to DINOv3 for instance segmentation

[ICLR 2026] The implementation of the paper Foundation Visual Encoders Are Secretly Few-Shot Anomaly Detectors

Adapting DINOv3 to downstream segmentation tasks

[OV-DEIM] Real-time DETR-Style Open-Vocabulary Object Detection with GridSynthetic Augmentation

Lightweight head for depth estimation using DINOv3 as backbone

unofficial JAX implementation of DINOv3, translated in full from the original Meta PyTroch reference implementation (Meta please don't sue me)

[CVPR 2026] Ultra-Low Bitrate Perceptual Image Compression with Shallow Encoder

A PyTorch implementation of an image classification system based on the DINOv3 (self-DIstillation with NO labels) vision transformer. This project provides a complete training pipeline with distributed data parallel (DDP) support, advanced data augmentation, and multiple loss functions including supervised contrastive learning.

Lightweight head for object detection using DINOv3 as backbone

This VL-JEPA implimentation takes direct insperation from the original VL-JEPA paper