Introducing YOLOBench by Deeplite!
A benchmark of over 900 YOLO-based object detectors
We are excited to release YOLOBench, a latency-accuracy benchmark of over 900 YOLO-based object detectors for embedded use cases on different edge hardware platforms. Accepted at the ICCV 2023 RCV workshop, you can read the full paper on arXiv.
- With a plethora of YOLO-based object detection models available, it can be overwhelming to choose the right one for a specific use case and hardware
- A comprehensive, fair and controlled comparison , latency-accuracy benchmark of over 900 models.
- Designed for embedded use cases on different edge hardware platforms
- Collects accuracy and latency number for each model and dataset combination across different hardware platforms
- Evaluates several zero-cost accuracy estimators that can predict the accuracy of a model without training it
- Check out the interactive
YOLOBench app on HuggingFace Spaces where you can find the best YOLO model for your edge device!
Questions? Contact us at yolobench@deeplite.ai!
