Enum ObjectDetectorScalingMode
- Assembly
- VisioForge.Core.dll
Specifies the scaling strategy for the search window used in object detection algorithms. This mode determines how the detector iterates through different sizes of objects to find matches.
public enum ObjectDetectorScalingModeFields
GreaterToSmaller = 0-
The object detector will start searching with larger windows and gradually scale down to smaller ones. This mode can be efficient if the target objects are expected to be large.
SmallerToGreater = 1-
The object detector will start searching with smaller windows and gradually scale up to larger ones. This mode can be efficient if the target objects are expected to be small.
Remarks
The choice of scaling mode can impact both the performance and accuracy of object detection. For example, if you expect to detect large objects primarily, starting with larger windows might be more efficient. GreaterToSmaller mode starts with large search windows and progressively reduces size, efficient for detecting large objects first. SmallerToGreater mode starts with small search windows and progressively increases size, efficient for detecting small objects first. The scaling strategy affects detection order and early termination possibilities in single-object search modes. When objects are primarily close to camera (large in frame), GreaterToSmaller finds them faster. When objects are primarily far from camera (small in frame), SmallerToGreater finds them faster. Cascade classifiers and sliding window detectors use these modes to optimize multi-scale object detection. Performance impact varies with frame resolution, object density, and detection algorithm characteristics. Optimal mode selection depends on application context: surveillance (often distant/small), kiosks (close/large), or variable scenarios (use heuristics).