Find the Leak, Fix the Split: Cluster-Based Method to Prevent Leakage in Video-Derived Datasets

2025/11/17 by Glazner, Noam, Tsfaty, Noam, Shalev, Sharon +1
Computer Science · #Advanced Malware Detection Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Security and Verification in Computing #User Authentication and Security Systems

paper · doi:10.48550/arxiv.2511.13944

openalex publication_date 2025/11/17 · openalex created_date 2025/11/20 · openalex updated_date 2026/07/28

Abstract

We propose a cluster-based frame selection strategy to mitigate information leakage in video-derived frames datasets. By grouping visually similar frames before splitting into training, validation, and test sets, the method produces more representative, balanced, and reliable dataset partitions.

Citations

• DINOv3
• XFeat: Accelerated Features for Lightweight Image Matching
• Sigmoid Loss for Language Image Pre-Training
• Learning Transferable Visual Models From Natural Language Supervision
• ImageNet Large Scale Visual Recognition Challenge
• ImageNet Large Scale Visual Recognition Challenge
• UCF101: A Dataset of 101 Human Actions Classes From Videos in The Wild
• Distinctive Image Features from Scale-Invariant Keypoints

Related

• Video Set Distillation: Information Diversification and Temporal Densification
• A novel learning-based frame pooling method for Event Detection
• VideoGuard: Protecting Video Content from Unauthorized Editing
• Distill Video Datasets into Images
• Scene Detection Policies and Keyframe Extraction Strategies for Large-Scale Video Analysis