Attention Guided Cosine Margin to Overcome Class-Imbalance in Few-Shot Road Object Detection

Few-Shot Object Detectors (FSOD) are tasked to localize and classify objects in an image given only a few data samples. Recent trends in FSOD research show the adoption of metric and meta-learning techniques, which are prone to catastrophic forgetting and class confusion. To overcome these pitfalls in metric learning based FSOD techniques, we introduce an Attention Guided Cosine Margin (AGCM) that facilitates the creation of tighter and well separated class-specific feature clusters in the classification head of the object detector. The Attentive Proposal Fusion (APF) module introduced in AGCM minimizes catastrophic forgetting by reducing the intra-class variance among co-occurring classes. At the same time, the Cosine Margin penalty in AGCM increases the angular margin between confusing classes to overcome the challenge of class confusion between already learned (base) and newly added (novel) classes. We conduct our experiments on the India Driving Dataset (IDD), which presents a real-world class-imbalanced setting alongside popular FSOD benchmark PASCAL-VOC. Our method outperforms existing approaches by up to 6.4 mAP points on the IDD-OS and up to 2.0 mAP points on the IDD-10 splits for the 10-shot setting. On the PASCAL-VOC dataset, we outperform existing approaches by up to 4.9 mAP points.