EfficientViT-SAM: Accelerated Segment Anything Model Without Accuracy Loss
This work addresses the high computational cost of SAM for users in computer vision, though it is incremental as it focuses on acceleration rather than new capabilities.
The paper tackles the computational inefficiency of the Segment Anything Model (SAM) by replacing its heavy image encoder with EfficientViT, achieving a 48.9x speedup on A100 GPU without accuracy loss.
We present EfficientViT-SAM, a new family of accelerated segment anything models. We retain SAM's lightweight prompt encoder and mask decoder while replacing the heavy image encoder with EfficientViT. For the training, we begin with the knowledge distillation from the SAM-ViT-H image encoder to EfficientViT. Subsequently, we conduct end-to-end training on the SA-1B dataset. Benefiting from EfficientViT's efficiency and capacity, EfficientViT-SAM delivers 48.9x measured TensorRT speedup on A100 GPU over SAM-ViT-H without sacrificing performance. Our code and pre-trained models are released at https://github.com/mit-han-lab/efficientvit.