AI Object Detection
Detect and label objects in any image using AI. Identifies 80+ object types with bounding boxes \u2014 runs entirely in your browser.
Processed on your device — never uploaded
Upload an Image
Your image never leaves your deviceWorks offline after first use
Frequently Asked Questions
How does AI object detection work?
This tool uses DETR (Detection Transformer), a deep learning model from Facebook Research that runs entirely in your browser via WebAssembly. It analyzes the image and predicts bounding boxes with labels for every object it recognizes.
What objects can it detect?
The model is trained on the COCO dataset and can detect 80+ object categories including people, vehicles (car, truck, bus, bicycle, motorcycle), animals (cat, dog, bird, horse), furniture (chair, couch, bed, table), electronics (laptop, phone, TV), food items (banana, apple, pizza, cake), and much more.
What does the confidence threshold do?
The confidence threshold controls how certain the model must be before showing a detection. Higher thresholds (e.g. 80%) only show very confident detections, reducing false positives. Lower thresholds (e.g. 20%) show more detections but may include some incorrect ones. The default 50% is a good balance.
Is my image uploaded to a server?
No. All processing happens entirely on your device, in your browser. Your images never leave your device. The AI model is downloaded once (~43 MB) and cached for future use, so it works offline after the first use.