# C-DIRA: Computationally Efficient Dynamic ROI Routing and Domain-Invariant Adversarial Learning for Lightweight Driver Behavior Recognition

arXiv: https://arxiv.org/abs/2512.08647 · HTML: https://mlsanity.com/paper/2512.08647/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 49 references that are in the map.
- First posted on arXiv 9 Dec 2025.

## Builds on (most cited first, 23 of 23)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications](https://mlsanity.com/paper/1704.04861/) (2017)
- [Searching for MobileNetV3](https://mlsanity.com/paper/1905.02244/) (2019)
- [RepVGG: Making VGG-style ConvNets Great Again](https://mlsanity.com/paper/2101.03697/) (2021)
- [Adversarial Examples Improve Image Recognition](https://mlsanity.com/paper/1911.09665/) (2019)
- [MambaOut: Do We Really Need Mamba for Vision?](https://mlsanity.com/paper/2405.07992/) (2024)
- [TResNet: High Performance GPU-Dedicated Architecture](https://mlsanity.com/paper/2003.13630/) (2020)
- [Fast Vision Transformers with HiLo Attention](https://mlsanity.com/paper/2205.13213/) (2022)
- [SwiftFormer: Efficient Additive Attention for Transformer-based Real-time Mobile Vision Applications](https://mlsanity.com/paper/2303.15446/) (2023)
- [MobileViTv3: Mobile-Friendly Vision Transformer with Simple and Effective Fusion of Local, Global and Input Features](https://mlsanity.com/paper/2209.15159/) (2022)
- [Towards Lightweight Transformer via Group-wise Transformation for Vision-and-Language Tasks](https://mlsanity.com/paper/2204.07780/) (2022)
- [Vision-Language Models can Identify Distracted Driver Behavior from Naturalistic Videos](https://mlsanity.com/paper/2306.10159/) (2023)
- [A comparative study of generative adversarial networks for image recognition algorithms based on deep learning and traditional methods](https://mlsanity.com/paper/2408.03568/) (2024)
- [Adversarial Transfer Learning for Cross-domain Visual Recognition](https://mlsanity.com/paper/1711.08904/) (2017)
- [(Safe) SMART Hands: Hand Activity Analysis and Distraction Alerts Using a Multi-Camera Framework](https://mlsanity.com/paper/2301.05838/) (2023)
- [Semi-Supervised Self-Growing Generative Adversarial Networks for Image Recognition](https://mlsanity.com/paper/1908.03850/) (2019)
- [Minimum Noticeable Difference based Adversarial Privacy Preserving Image Generation](https://mlsanity.com/paper/2206.08638/) (2022)
- [Adversarial Learning of Hard Positives for Place Recognition](https://mlsanity.com/paper/2205.03871/) (2022)
- [Driver-Net: Multi-Camera Fusion for Assessing Driver Take-Over Readiness in Automated Vehicles](https://mlsanity.com/paper/2507.04139/) (2025)
- [Detecting driver distraction using stimuli-response EEG analysis](https://mlsanity.com/paper/1904.09100/) (2019)
- [Adversarial Masking Contrastive Learning for vein recognition](https://mlsanity.com/paper/2401.08079/) (2024)

## Abstract (arXiv)

Driver distraction behavior recognition using in-vehicle cameras demands real-time inference on edge devices. However, lightweight models often fail to capture fine-grained behavioral cues, resulting in reduced performance on unseen drivers or under varying conditions. ROI-based methods also increase computational cost, making it difficult to balance efficiency and accuracy. This work addresses the need for a lightweight architecture that overcomes these constraints. We propose Computationally efficient Dynamic region of Interest Routing and domain-invariant Adversarial learning for lightweight driver behavior recognition (C-DIRA). The framework combines saliency-driven Top-K ROI pooling and fused classification for local feature extraction and integration. Dynamic ROI routing enables selective computation by applying ROI inference only to high difficulty data samples. Moreover, pseudo-domain labeling and adversarial learning are used to learn domain-invariant features robust to driver and background variation. Experiments on the State Farm Distracted Driver Detection Dataset show that C-DIRA maintains high accuracy with significantly fewer FLOPs and lower latency than prior lightweight models. It also demonstrates robustness under visual degradation such as blur and low-light, and stable performance across unseen domains. These results confirm C-DIRA's effectiveness in achieving compactness, efficiency, and generalization.