IC-006The Retriever-Dictionary module improves object detection accuracy of YOLOv7, YOLOv9, Faster R-CNN, and Deformable DETR on COCO 2017

Hao-Tang Tsui, Chien-Yao Wang, Hong-Yuan Mark Liao

SourceYOLO-RD: Introducing Relevant and Compact Explicit Knowledge to YOLO by Retriever-Dictionary

The paper introduces a Retriever-Dictionary (RD) module that enhances the performance of several pre-existing object detection models by enabling them to leverage external dataset knowledge during training. The module was integrated into YOLOv7, YOLOv9, Faster R-CNN, and Deformable DETR, with the dictionary initialized using different encoders (a vision model, a vision-language model, and a large language model). The primary finding is that incorporating this module consistently improves the mean Average Precision (mAP) of these models on the COCO 2017 validation set, with less than a 1% increase in model parameters for YOLO-based models.

Evidence
correlational
Key metric
YOLOv9 with VLM initialization: 53.36% mAP (↑1.37%), 70.55% mAP@.5 (↑1.43%); YOLOv7 with VLM initialization: 51.75% mAP (↑3.42%), 70.12% mAP@.5 (↑2.15%); Faster R-CNN with VLM initialization: 40.50% mAP (↑5.47%), 60.40% mAP@.5 (↑2.37%); Deformable DETR with VLM initialization: 44.40% mAP (↑1.37%), 63.30% mAP@.5 (↑1.12%)
Caveat
The results for YOLOv7 and YOLOv9 are slightly lower than originally reported in their respective papers, which the authors attribute to differences in the number of GPUs used and the reduced effectiveness of batch normalization with smaller batch sizes.
Model
YOLOv7, YOLOv9, Faster R-CNN / Faster R-CNN R50 / Faster R-CNN X101, Deformable DETR
Datasets
MS COCO / COCO / COCO 2014 / COCO 2017 / COCO 20k / COCO-it / COCO-wl [eval]
Methods
Retriever-Dictionary module / Retriever-Dictionary (RD) module / RD module [primary], Knowledge Distillation [compared-to], RALF [compared-to]
Related work
YOLOv7 [builds-on], YOLOv9 [builds-on], Faster R-CNN / Faster R-CNN R50 / Faster R-CNN X101 [builds-on], Deformable DETR [builds-on], CLIP / CLIP-ViT (LC) [context]
Extraction
automatic-extraction