NTIRE 2025 Challenge on Cross-Domain Few-Shot Object Detection: Methods and Results

2025-04-14

AI Summary

This work addresses Cross-Domain Few-Shot Object Detection (CD-FSOD), a challenging task requiring robust generalization from minimal target-domain annotations. We propose a unified framework integrating meta-learning, feature alignment, domain adaptation, and prompt-based fine-tuning, augmented with multi-scale feature distillation and dynamic pseudo-label optimization to enhance cross-domain generalization under extreme label scarcity. To foster community advancement, we launch the first international CD-FSOD Challenge, establishing an open/closed-track evaluation paradigm to standardize benchmarking. The challenge attracted 152 registered teams, with 42 submitting valid entries; all 13 finalist methods surpassed the baseline, achieving substantial mAP gains—e.g., on PASCAL→COCO transfers—thereby yielding multiple new state-of-the-art approaches. Our work provides both conceptual advances in few-shot cross-domain transfer and a rigorous, publicly accessible benchmark platform for future research.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Object Detection & Categorization
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Cross-Domain Few-Shot Object Detection (CD-FSOD) poses significant challenges to existing object detection and few-shot detection models when applied across domains. In conjunction with NTIRE 2025, we organized the 1st CD-FSOD Challenge, aiming to advance the performance of current object detectors on entirely novel target domains with only limited labeled data. The challenge attracted 152 registered participants, received submissions from 42 teams, and concluded with 13 teams making valid final submissions. Participants approached the task from diverse perspectives, proposing novel models that achieved new state-of-the-art (SOTA) results under both open-source and closed-source settings. In this report, we present an overview of the 1st NTIRE 2025 CD-FSOD Challenge, highlighting the proposed solutions and summarizing the results submitted by the participants.

Problem

• Advancing object detection in novel domains with limited data
• Addressing challenges in cross-domain few-shot object detection
• Evaluating new models for state-of-the-art performance across domains

Innovation

• Cross-Domain Few-Shot Object Detection challenge
• Novel models for state-of-the-art results
• Limited labeled data in new domains