Symbiosis-Inspired Knowledge Distillation for Incremental Object Detection

2026-07-15

AI Summary

This work addresses the challenge of catastrophic forgetting and performance degradation in incremental object detection, which often arises from neglecting object co-occurrence relationships. To mitigate shared representation distortion and confusion between old and new classes, the study introduces object co-occurrence into this task for the first time and proposes a co-occurrence-inspired knowledge distillation framework comprising two complementary mechanisms: Spatial Co-occurrence Distillation (SpSD) and Semantic Co-occurrence Distillation (SeSD). SpSD refines evidence in highly overlapping regions through slot alignment, while SeSD preserves inter-class semantic structure by constructing confidence-weighted prototypes and applying soft ranking. Evaluated on multiple benchmarks, the proposed method significantly outperforms existing approaches, effectively alleviating catastrophic forgetting and simultaneously enhancing detection performance for both old and new categories.

Abstract

Incremental object detection (IOD) aims to extend detectors to new categories while retaining previously acquired knowledge. Existing methods often adopt a class incremental learning perspective, separating feature spaces to sharpen decision boundaries. However, this separation-oriented paradigm may overlook object symbiosis in detection, where co-occurrence and occlusion introduce spatial and semantic dependencies that benefit from shared representations. Ignoring these dependencies distorts the shared representations, exacerbates confusion between old and new classes, and accelerates catastrophic forgetting. To address this, we propose Symbiosis-Inspired Knowledge Distillation (SIKD), which explicitly leverages object symbiosis at two complementary levels. Spatial Symbiosis Distillation (SpSD) focuses on symbiotic regions where the old model responds with high overlap to objects in the new task. It preserves generalizable old class cues, suppresses class-specific bias and redundancy, and distills the refined evidence to the new model at matched spatial locations with slot-aligned supervision. Semantic Symbiosis Distillation (SeSD) maintains class level structure by forming confidence weighted prototypes for old classes and aligning their inter class soft ranks over the old class logits, which stabilizes the semantic topology during adaptation. Extensive experiments demonstrate the effectiveness and superiority of the proposed method.

Problem

• Incremental Object Detection
• Object Symbiosis
• Catastrophic Forgetting
• Knowledge Distillation
• Class Incremental Learning

Innovation

• Symbiosis-Inspired Knowledge Distillation
• Incremental Object Detection
• Spatial Symbiosis Distillation
• Semantic Symbiosis Distillation
• Catastrophic Forgetting