Highlighting Object Category Immunity for the Generalization of Human-Object Interaction Detection
Human-Object Interaction (HOI) detection plays a core role in activity understanding. As a compositional learning problem (human-verb-object), studying its generalization matters. However, widely-used metric mean average precision (mAP) fails to model the compositional generalization well. Thus, we propose a novel metric, mPD (mean Performance Degradation), as a complementary of mAP to evaluate the performance gap among compositions of different objects and the same verb. Surprisingly, mPD reveals that previous methods usually generalize poorly. With mPD as a cue, we propose Object Category (OC) Immunity to boost HOI generalization. The idea is to prevent model from learning spurious object-verb correlations as a short-cut to over-fit the train set. To achieve OC-immunity, we propose an OC-immune network that decouples the inputs from OC, extracts OC-immune representations, and leverages uncertainty quantification to generalize to unseen objects. In both conventional and zero-shot experiments, our method achieves decent improvements. To fully evaluate the generalization, we design a new and more difficult benchmark, on which we present significant advantage. The code is available at https://github.com/Foruck/OC-Immunity.
Code (1)
Tasks
Human-Object Interaction DetectionObjectUncertainty QuantificationSimilar Papers 제목 키워드 기반
DVF: Advancing Robust and Accurate Fine-Grained Image Retrieval with Retrieval Guidelines
Fine-grained image retrieval (FGIR) is to learn visual representations that distinguish visually similar objects while maintaining generalization. Existing methods propose to generate discriminative features, but rarely …
Image RetrievalRetrievalA new oscillatory regime in two-strain epidemic models with partial cross-immunity
Infectious diseases often involve multiple strains that interact through the immune response generated after an infection. This study investigates the conditions under which a two-strain epidemic model with partial cross…
Object-Category Aware Reinforcement Learning
Object-oriented reinforcement learning (OORL) is a promising way to improve the sample efficiency and generalization ability over standard RL. Recent works that try to solve OORL tasks without additional feature engineer…
Feature EngineeringObjectObject Discoveryreinforcement-learning+2Bifurcations and multi-stability in a model of cytokine-mediated autoimmunity
This paper investigates the dynamics of immune response and autoimmunity with particular emphasis on the role of regulatory T cells (Tregs), T cells with different activation thresholds, and cytokines in mediating T cell…
Self-organizing maps and generalization: an algorithmic description of Numerosity and Variability Effects
Category, or property generalization is a central function in the human cognition. It plays a crucial role in a variety of domains, such as learning, everyday reasoning, specialized reasoning, and decision making. Judgin…
Decision Making