APOC Usage

Step-by-Step Protocol for Using Trackastra

Using APOC: 

Object and semantic segmentation: 

1. Access the object segmentation tool:

• Go to Tools > Segmentation/labeling > Object segmentation(APOC).

2. Load necessary layers: 

• Image layer: the input image. 

• Labels layer: your annotations (ground truth).

3. Select training data:

• Choose the image (or channels) you want to use for training the classifier.

4. Annotate classes:

• Label 1: background.

• Label 2: foreground.

5. Select labels:

• Choose the labels layer with your annotated regions. 

• These annotations will be used as training data (ground truth) for the classifier.

6. Select features: 

• Choose image features for training. 

7. Train the classifier:

• Click Train.

• The model will learn from your annotations.

8. Save the trained classifier: 

• Save the trained classifier as a .cl file. 

• This file is required to apply the model later to new images.

Applying a pretrained model:

1. Open the prediction tool:

• Go to Tools > Segmentation/labeling > Object Segmentation (Apply pretrained APOC).

2. Load data:

• Select the image layer you want to segment.

• Select the saved .cl classifier file.

3. Run prediction:

• Click Apply classifier/Predict segmentation

• A new labels layer will be generated.

4. Save results: 

• Save the segmentation output as a TIFF file. 

Probability maps:

1. Access the probability mapper tool: 

• Go to Tools > Filtering > Probability Mapper (APOC).

2. Annotate classes:

• Label 1: background

• Label 2: foreground

• Label 3: edges

3. Generate the probability map:

• The result will be an intensity image where pixel intensities represent the probability of belonging to each class (ranging from 0 to 1).