Sample-aware RandAugment: Search-free Automatic Data Augmentation for Effective Image Recognition

2025-08-11

AI Summary

Mainstream automated data augmentation (AutoDA) suffers from two key bottlenecks: prohibitively high search overhead and poor adaptability of augmentation policies across samples. To address these, we propose Sample-Responsive Augmentation (SRA), a search-free, sample-aware AutoDA method. SRA employs a lightweight sample complexity scoring module to dynamically assess image difficulty and accordingly schedules asymmetric augmentation strength—within the RandAugment framework—enabling zero-cost policy generation with high sample-level adaptivity. Crucially, SRA shifts the augmentation decision paradigm from expensive architecture- or policy-space search to efficient, per-sample dynamic response. This design drastically reduces computational overhead while improving generalization. On ImageNet, SRA achieves 78.31% Top-1 accuracy with ResNet-50. Moreover, models pretrained with SRA consistently outperform baselines when transferred to downstream object detection tasks, demonstrating both effectiveness and broad applicability.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Automatic data augmentation (AutoDA) plays an important role in enhancing the generalization of neural networks. However, mainstream AutoDA methods often encounter two challenges: either the search process is excessively time-consuming, hindering practical application, or the performance is suboptimal due to insufficient policy adaptation during training. To address these issues, we propose Sample-aware RandAugment (SRA), an asymmetric, search-free AutoDA method that dynamically adjusts augmentation policies while maintaining straightforward implementation. SRA incorporates a heuristic scoring module that evaluates the complexity of the original training data, enabling the application of tailored augmentations for each sample. Additionally, an asymmetric augmentation strategy is employed to maximize the potential of this scoring module. In multiple experimental settings, SRA narrows the performance gap between search-based and search-free AutoDA methods, achieving a state-of-the-art Top-1 accuracy of 78.31% on ImageNet with ResNet-50. Notably, SRA demonstrates good compatibility with existing augmentation pipelines and solid generalization across new tasks, without requiring hyperparameter tuning. The pretrained models leveraging SRA also enhance recognition in downstream object detection tasks. SRA represents a promising step towards simpler, more effective, and practical AutoDA designs applicable to a variety of future tasks. Our code is available at href{https://github.com/ainieli/Sample-awareRandAugment}{https://github.com/ainieli/Sample-awareRandAugment

Problem

• Improves AutoDA by avoiding time-consuming search processes
• Enhances policy adaptation for better image recognition performance
• Dynamically adjusts augmentations per sample for optimal results

Innovation

• Search-free dynamic augmentation policy adjustment
• Heuristic scoring for sample complexity evaluation
• Asymmetric augmentation strategy for performance optimization