SimpleAICV provides lightweight, easy-to-use training pipelines for various computer vision tasks. It supports object detection models like FCOS, RetinaNet, YOLACT, and SOLOv2, segmentation models including SAM and SAM2, and vision transformers such as ViT, DINOv3, and ConvFormer. The repository includes Gradio demos, inference examples, and training scripts for datasets like COCO, ImageNet, VOC, and ADE20K.