Object Detection

Object detection is a computer vision task that simultaneously identifies what objects are

in an image and where they are located. This tool uses YOLOS-tiny, a transformer-based

model trained on the COCO dataset, to detect 80 categories of everyday objects and return

labeled bounding boxes with confidence scores for each detection.

80 detectable object classes span everyday life: people, bicycles, cars, buses, trucks,

motorcycles, traffic lights, animals (cat, dog, bird, horse, cow, sheep, elephant, bear,

zebra), furniture (chair, sofa, bed, dining table), electronics (laptop, phone, keyboard,

mouse, TV, microwave, oven), and kitchen and food items.

The confidence slider lets you tune results in real time — raise it to show only high-certainty

detections, lower it to catch uncertain ones. Toggle entire object classes on or off to focus

on specific categories. Export the annotated image as PNG or download all detection coordinates,

labels, and confidence scores as JSON for integration with other tools or code.

Common use cases: analyzing photos for cataloguing and inventory, counting objects in research

images, testing computer vision pipelines without writing code, verifying accessibility

descriptions, security analysis, and educational exploration of how object detection works.

The 25MB model downloads once and caches for instant reuse. All processing is local —

your images never leave your device.