YOLOS-S — Object Detection AI Model
YOLOS-Small is a pure-transformer detector over the same 91 COCO categories as DETR, smaller and quicker at a little cost in accuracy. Take it when throughput matters more than catching the last object. CPU-capable.
What it does
YOLOS-Small finds the objects in a photograph and returns each one with a bounding box and a confidence score, over the same 91-category COCO label space as DETR.
The name stands for You Only Look at One Sequence: a pure transformer with no convolutional backbone, treating detection as sequence prediction over image patches. Smaller and quicker than DETR ResNet-50 at a little cost in accuracy, and it will run on CPU.
When to use it
- General detection across the COCO categories, with no training
- Throughput first, when DETR ResNet-50 is more wait than needed
- Several objects located in one picture in a single pass
- Running on CPU instead of occupying a GPU
- Boxes plus a text list of the findings, for a later step to read
Details
- CategoryObject Detection
- Year2021
- LicenseApache-2.0
Compliance & Provenance
- ProviderOpen-source · Specialized
- EU AI Act RiskMinimal Risk
- Art. 50 TransparencyNot applicable
Inputs & Outputs
- ImageInputimage
Source image for object detection
- ImageOutputimage
Detected objects are highlighted with bounding boxes
- TextOutputstring
Detected types of objects with confidence scores
Tags
Alternatives in Object Detection
- DETR
General object detection on 91 COCO classes (DETR).
- DETR (Face Detector)
Face detection (DETR ResNet-50 fine-tuned). Fast.
- DETR (License Plate Detector)
License-plate detection (DETR ResNet-50). Fast.
- RF-DETR Medium
Real-time object detection on 80 COCO classes (RF-DETR Medium).
- YOLO-S (Fashion Item Detection)
Fashion item detection — 46 apparel classes (YOLOS).