Figure 1.

The process of particle picking with CryoSegNet. (A) An input micrograph is first denoised and then sent to the U-net model. (B) U-Net model outputs a segmentation mask for each micrograph that is fed to SAM automatic mask generator for predicting the bounding boxes of protein particles. (C) The output generated by SAM is further processed based on thresholding the prediction confidence scores to filter out some false particles to generate the final output of picked particles stored in .star files.