tanırım
AI & Model

YOLO11

ModelObject DetectionPyTorchONNXBrowser

An object detection model from the YOLO family, released by Ultralytics in 2024. It processes the image in a single pass and produces a box and class scores for every candidate. Extra boxes for the same object are removed with NMS. It comes in five sizes, from n to x, and can also run in the browser once exported to ONNX.

General Properties

Architecture / Approach
Single-stage, anchor-free object detection
Trained On
COCO
Model Size
~38 MB (s size, ONNX)
Input Resolution
640 × 640 (default)
License
AGPL-3.0
// Help

Frequently Asked Questions

It is used to find objects in images and mark them with a box and a class. The model is retrained on the dataset that is needed, exported to ONNX and run in the visitor's browser.

The image is scaled down to the input size before it reaches the model, which reduces the detail of small objects. A larger input size finds small objects better, but the model runs slower. For example, on the VisDrone val set the model trained at 1280 raises mAP50 from 0.347 at 640 to 0.503.

AGPL-3.0. Model weights trained with it fall under AGPL-3.0 as well.