Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Mining Instance-Centric Vision-Language Contexts for Human-Object Interaction Detection

Soo Won Seo, KyungChae Lee, Hyungchan Cho, Taein Son, Nam Ik Cho, Jun Won Choi

Abstract

Human-Object Interaction (HOI) detection aims to localize human-object pairs and classify their interactions from a single image, a task that demands strong visual understanding and nuanced contextual reasoning. Recent approaches have leveraged Vision-Language Models (VLMs) to introduce semantic priors, significantly improving HOI detection performance. However, existing methods often fail to fully capitalize on the diverse contextual cues distributed across the entire scene. To overcome these limitations, we propose the Instance-centric Context Mining Network (InCoM-Net)-a novel framework that effectively integrates rich semantic knowledge extracted from VLMs with instance-specific features produced by an object detector. This design enables deeper interaction reasoning by modeling relationships not only within each detected instance but also across instances and their surrounding scene context. InCoM-Net comprises two core components: Instancecentric Context Refinement (ICR), which separately extracts intra-instance, inter-instance, and global contextual cues from VLM-derived features, and Progressive Context Aggregation (ProCA), which iteratively fuses these multicontext features with instance-level detector features to support high-level HOI reasoning. Extensive experiments on the HICO-DET and V-COCO benchmarks show that InCoM-Net achieves state-of-the-art performance, surpassing previous HOI detection methods. Code is available at https://github.com/nowuss/InCoM-Net.

Mining Instance-Centric Vision-Language Contexts for Human-Object Interaction Detection

Abstract

Human-Object Interaction (HOI) detection aims to localize human-object pairs and classify their interactions from a single image, a task that demands strong visual understanding and nuanced contextual reasoning. Recent approaches have leveraged Vision-Language Models (VLMs) to introduce semantic priors, significantly improving HOI detection performance. However, existing methods often fail to fully capitalize on the diverse contextual cues distributed across the entire scene. To overcome these limitations, we propose the Instance-centric Context Mining Network (InCoM-Net)-a novel framework that effectively integrates rich semantic knowledge extracted from VLMs with instance-specific features produced by an object detector. This design enables deeper interaction reasoning by modeling relationships not only within each detected instance but also across instances and their surrounding scene context. InCoM-Net comprises two core components: Instancecentric Context Refinement (ICR), which separately extracts intra-instance, inter-instance, and global contextual cues from VLM-derived features, and Progressive Context Aggregation (ProCA), which iteratively fuses these multicontext features with instance-level detector features to support high-level HOI reasoning. Extensive experiments on the HICO-DET and V-COCO benchmarks show that InCoM-Net achieves state-of-the-art performance, surpassing previous HOI detection methods. Code is available at https://github.com/nowuss/InCoM-Net.

Paper Structure

This paper contains 31 sections, 8 equations, 5 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Human-Object Interaction Detection
  4. VLM Integration for HOI Detection
  5. Method
  6. Overview
  7. Instance-centric Context Mining
  8. Instance-centric Context Refinement (ICR).
  9. Progressive Context Aggregation (ProCA).
  10. HO Pair Generation and Interaction Reasoning
  11. HO Pair Generator.
  12. Interaction Decoder.
  13. Masked Feature Training (MFT)
  14. Experiments
  15. Datasets and Evaluation Settings
  16. ...and 16 more sections

Figures (5)

  • Figure 1: Illustration of three levels of contextual information. For each instance, contextual information can be distinguished into intra-instance, inter-instance, and global contexts, each providing complementary cues for interpreting human–object interactions.
  • Figure 2: Overall framework of InCoM-Net. Left: overview of Instance-centric Context Mining that integrates multi-context VLM features with instance-level features. Right: masked feature training (MFT), balancing the utilization of heterogeneous feature sources via masking.
  • Figure 3: Structure of ICR. The illustration shows the process of generating the multi-context features for the $i$-th instance.
  • Figure 4: Structure of ProCA.
  • Figure 5: Visualization of activation maps from the interaction decoder.