Table 1 Overview on the image processing tasks implemented in InstantDL, required input, label, and output format
From: InstantDL: an easy-to-use deep learning pipeline for image segmentation and classification
| Â | Semantic segmentation | Instance segmentation | Pixel-wise regression | Classification |
|---|---|---|---|---|
Input image | 2D & 3D | 2D | 2D & 3D | 2D |
Labels | Binary images | Images with float pixel values | Images with float pixel values | Labels in a .csv file |
Output | Binary images | Binary masks | Images with float pixel values | Labels in a .csv file |
Architecture | U-Net | Mask RCNN | U-Net | ResNet50 |