SAM2-Unet is a hybrid segmentation model integrating the Segment Anything Model (SAM) with U-Net, optimized for medical image segmentation and few-shot learning. It incorporates SAM's visual prompt mechanism into U-Net's encoder-decoder structure, enabling dynamic target guidance via interactive point/box inputs while retaining skip connections for multi-scale feature fusion. Lightweight adapters fine-tune SAM's pretrained weights to enhance sensitivity to low-contrast regions in medical images (e.g., CT/MRI) and reduce reliance on large annotated datasets. Supporting zero-shot transfer and few-shot tuning, it improves Dice scores by ~8% over traditional U-Net on BraTS and ISIC benchmarks with low computational overhead, ideal for clinical diagnostics and real-time lesion localization.

Source model

• Input shape: 1x3x352x352
• Number of parameters: 206.51M
• Model size: 849.34M
• Output shape: 1x1x352x352

The source model can be found here

When the user has fine-tuned the source model, the model conversion process must be performed again.

Users can refer to either of the following two methods to complete the model conversion:

• Using AIMO for model conversion: Click Model Conversion Reference in the Performance Reference section on the right to view the conversion steps.

• Using Qualcomm QNN for model conversion: Please refer to the Qualcomm QNN Documentation.

The model performance benchmarks and example code provided by Model Farm are all implemented based on the APLUX AidLite SDK.

For models in .bin format, you can use either of the following two inference engines to run inference on Qualcomm chips:

Inference using APLUX AidLite: please refer to the APLUX AidLite Developer Documentation

Inference using Qualcomm QNN: Please refer to the Qualcomm QNN Documentation

Inference Example Code

The inference example code is implemented using the AidLite SDK.

Click Model & Code to download the model files and the inference code package. The file structure is as follows:

/model_farm_{model_name}_aidlite
    
    |__ models # folder where model files are stored

    |__ python # aidlite python model inference example

    |__ cpp # aidlite cpp model inference example

    |__ README.md