Stability Plasticity Decoupled Fine-tuning For Few-shot end-to-end Object Detection

Few-shot object detection(FSOD) aims to design methods to adapt object detectors efficiently with only few annotated samples. Fine-tuning has been shown to be an effective and practical approach. However, previous works often take the classical base-novel two stage fine-tuning procedure but ignore the implicit stability-plasticity contradiction among different modules. Specifically, the random re-initialized classifiers need more plasticity to adapt to novel samples. The other modules inheriting pre-trained weights demand more stability to reserve their class-agnostic knowledge. Regular fine-tuning which couples the optimization of these two parts hurts the model generalization in FSOD scenarios. In this paper, we find that this problem is prominent in the end-to-end object detector Sparse R-CNN for its multi-classifier cascaded architecture. We propose to mitigate this contradiction by a new three-stage fine-tuning procedure by introducing an addtional plasticity classifier fine-tuning(PCF) stage. We further design the multi-source ensemble(ME) technique to enhance the generalization of the model in the final fine-tuning stage. Extensive experiments verify that our method is effective in regularizing Sparse R-CNN, outperforming previous methods in the FSOD benchmark.