MIC-Learner

MIC-Learner is a plugin for ImageJ that provides a GUI for discovering AI approaches in Image Processing.

The GUI divides into 2 main parts:

• the first in task-driven to help the user to find the best approach for a given task.
• the second is algorithm-driven to help the user to find the best algorithm for a given task.

The tasks are divided into 4 main categories:

• Image Classification
• Object Detection
• Image Segmentation (semantic or instance segmentation)
• Image Enhancement (denoising)

With each time a description of what are the expected results for each type of task.

The provided algorithms corresponds to classics in computer vision.

• Random Forest (using) Weka
• Multi-Layered Perceptron (MLP)
• Convolutional Neural Network (CNN)
• You Only Look Once (YOLO)
• U-Net

The algorithms are explained and a test model is provided to allow the user to test the performance of the algorithms. The test models use-cases are explained along with the algorithms. Tests can be performed on a sample image or on an image provided by the user.