Browser Vest Detector

This demo runs the exported YOLO vest detector directly in the browser. Upload an image or use one of the bundled test samples.

Back to course
Or pick a bundled test image 10 sample photos

Live comparison

? Help

Detection thresholds

Confidence (0.33) is the minimum score a detection must have before it is drawn. At 0.33, the simulator only shows boxes whose model confidence is 33% or higher. Raising it makes the detector stricter (fewer false positives, but more missed vests). Lowering it makes it more permissive (more detections, but also more noise).

IoU NMS (0.41) is the overlap threshold used by Non-Maximum Suppression. If two boxes overlap too much, and that overlap is above 0.41, the weaker box is removed so you do not keep duplicate detections for the same vest. Lower values remove overlapping boxes more aggressively; higher values keep more nearby boxes.

Input image vs detected output

Select one of the sample images or upload your own file, then press the arrow to run the model and compare both views side by side.

Image to infer
Selected input image preview
prueba1.png
Loading ONNX Runtime and model metadata...
Inference result
Click the arrow to generate detections.