Semi-supervised Object Detection via Virtual Category Learning

7 Jul 2022  ·  Changrui Chen, Kurt Debattista, Jungong Han ·

Due to the costliness of labelled data in real-world applications, semi-supervised object detectors, underpinned by pseudo labelling, are appealing. However, handling confusing samples is nontrivial: discarding valuable confusing samples would compromise the model generalisation while using them for training would exacerbate the confirmation bias issue caused by inevitable mislabelling. To solve this problem, this paper proposes to use confusing samples proactively without label correction. Specifically, a virtual category (VC) is assigned to each confusing sample such that they can safely contribute to the model optimisation even without a concrete label. It is attributed to specifying the embedding distance between the training sample and the virtual category as the lower bound of the inter-class distance. Moreover, we also modify the localisation loss to allow high-quality boundaries for location regression. Extensive experiments demonstrate that the proposed VC learning significantly surpasses the state-of-the-art, especially with small amounts of available labels.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Semi-Supervised Object Detection COCO 0.5% labeled data VC mAP 19.46 # 3
Semi-Supervised Object Detection COCO 10% labeled data VC mAP 34.82 # 14
detector FasterRCNN-Res50 # 1
Semi-Supervised Object Detection COCO 1% labeled data VC mAP 23.86 # 8
Semi-Supervised Object Detection COCO 2% labeled data VC mAP 27.70 # 10
Semi-Supervised Object Detection COCO 5% labeled data VC mAP 32.05 # 11

Methods


No methods listed for this paper. Add relevant methods here