<aside> 🛠

Generate Masks for Detected Objects

Creates precise binary masks for each detected object class from object detection model output. Automatically generate masks based on bounding boxes and class labels. Perfect for selective processing, class-specific masking, batch mask generation, and validation workflows. Works seamlessly with any object detection model (YOLO, YOLOv11n-Face-Detection, etc.)..

</aside>

What it does

Image Masker by Class takes output from an Object Detection model (bounding boxes and class labels) and automatically creates binary masks for each detected object. Select which object classes to mask, and the tool generates separate or combined masks highlighting exactly those regions. Perfect for isolating specific object types for further AI processing, inpainting, segmentation, or analysis. Works with any detection model output.

Problem it solves

• Selective class masking – Create masks only for specific object classes (e.g., just "person" or "car")
• Class-specific processing – Direct AI models to process only certain object types
• Detection validation – Visually inspect detected regions by converting to masks
• Mask from detection output – Skip manual masking—use automatic detection results instead
• Multi-class workflows – Generate separate masks for each class or combine related classes
• Inpainting preparation – Create masks for detected objects ready for removal/reconstruction
• Segmentation refinement – Use detection masks as starting point for segmentation
• Quality control – Verify detection accuracy by reviewing generated masks

Input/Output

<aside>

• Input: Images and Text
  • Images
    • Object Detection model output – Bounding boxes [x_min, y_min, x_max, y_max] with class labels and confidence scores
      • Source: Any object detection model (YOLO, YOLOv11n-Face-Detection, Faster R-CNN, etc.)
      • Format: Detection coordinates + class names
    • Original image – RGB or RGBA image to mask
  • Text
    • Class selection - Which object classes to create masks for
      • Input: Class names (e.g., "person", "face", "car")
      • Format: Comma-separated list or multi-select from detected classes

• Output: Binary mask(s)
  • Format: Black/white binary mask (PNG with alpha channel)
  • Resolution: Same as input image </aside>

Features - Three Controls for Masking

<aside>

• Class Selection: Choose which semantic class to extract
  • Available classes: Person, Car, Sky, Building, Tree, Grass, Water, Dog, Cat, Bicycle, etc.
  • How it works: Browse detected/segmented classes, select one class to extract
  • Purpose: Specify exactly which semantic region you want

• Dilation of the Bounding Box: Expand or contract bounding boxes before creating masks
  • Range: Positive (expand outward) or Zero (exact)
  • Measured in: Pixels on all sides
  • How it works:
    • +20px dilation: Adds 20 pixels padding around detected region
    • 10px dilation: Removes 10 pixels from edges of detected region
    • 0 dilation: Uses exact detection boundary

• Handling Duplicated Classes (Multiple Instances): When multiple instances of same class exist, decide which to keep

  • Seven Options:

  Option	Behavior	Use Case
  individuals	Keep each instance separately	Extract all people/cars in frame seperately
  merge	Combine all instances into ONE bounding box	Extract all people/cars in frame together
  left most	Keep instance furthest LEFT	Extract leftmost person
  top most	Keep instance furthest TOP	Extract topmost object
  right most	Keep instance furthest RIGHT	Extract rightmost person
  bottom most	Keep instance furthest BOTTOM	Extract bottommost object
  center most	Keep instance closest to CENTER	Extract main subject in middle

</aside>

How to Use

<aside>

Image masker by class.mp4

1. Load detection → Get output from object detection model (DETR, YOLOS, etc.)
2. Select class → Add Image Masker by Class and choose target class from detected objects
3. Run Flow → Execute to generate masks </aside>