hjLabs AI Playground
Starting the on-device runtime…
DETR-ResNet-50 gives accurate boxes over the 80 COCO classes; YOLOS-tiny and YOLOv10n are far smaller and fast enough to run on a webcam stream. Boxes, labels and confidence scores are drawn over the image and exportable as JSON, which makes this a quick way to sanity-check a dataset or prototype a detection idea before committing to a Python pipeline. Nothing is uploaded, so you can point it at internal photos.
object detection onlinedetr demoyolo in browserfree object detectorwebcam object detectioncoco detection demobounding box tool aiwebgpu object detection
Not these models — they are fixed to their 80 training classes. For arbitrary categories described in words, use zero-shot object detection with OWL-ViT instead.
Yes with the tiny models. YOLOS-tiny and YOLOv10n reach usable frame rates on a discrete GPU; DETR-ResNet-50 is better suited to single images.
No. The model weights are downloaded from Hugging Face to your browser once, and every inference after that runs on your own GPU through WebGPU. Your text, images and audio are never sent anywhere — hjLabs.in has no inference server and no way to see your input.
Small models do. Recent mobile Chrome and Safari expose WebGPU, but memory is the binding limit — stay under roughly 200 MB of weights on Android, and expect iOS to reclaim the tab sooner. The catalog shows every download size up front so you can pick something that fits.
That depends on the individual model, not on this site. Each model page links to its Hugging Face repository, where the licence is stated — most of the catalog is Apache-2.0 or MIT, a few are research-only. We add nothing on top: hjLabs.in claims no rights over anything you produce here.