YOLO9000: Better, Faster, Stronger

TL;DR

YOLO9000 is a real-time object detection system capable of identifying over 9000 categories, combining improved detection accuracy with fast processing speeds and joint training on detection and classification datasets.

Contribution

The paper introduces YOLO9000, a novel real-time detection system that jointly trains on detection and classification data to recognize thousands of object categories.

Findings

YOLOv2 achieves 76.8 mAP at 67 FPS on VOC 2007.

YOLO9000 detects over 9000 object categories in real-time.

Joint training improves detection on classes without detection data.

Abstract

We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories. First we propose various improvements to the YOLO detection method, both novel and drawn from prior work. The improved model, YOLOv2, is state-of-the-art on standard detection tasks like PASCAL VOC and COCO. At 67 FPS, YOLOv2 gets 76.8 mAP on VOC 2007. At 40 FPS, YOLOv2 gets 78.6 mAP, outperforming state-of-the-art methods like Faster RCNN with ResNet and SSD while still running significantly faster. Finally we propose a method to jointly train on object detection and classification. Using this method we train YOLO9000 simultaneously on the COCO detection dataset and the ImageNet classification dataset. Our joint training allows YOLO9000 to predict detections for object classes that don't have labelled detection data. We validate our approach on the ImageNet detection task. YOLO9000 gets 19.7 mAP on the ImageNet detection validation set despite only having detection data for 44 of the 200 classes. On the 156 classes not in COCO, YOLO9000 gets 16.0 mAP. But YOLO can detect more than just 200 classes; it predicts detections for more than 9000 different object categories. And it still runs in real-time.