AI Quantifies Dermal Exposure from Images for Safety

Hua Qian, Manisha Kotha, Tuan Tran, Jennifer Shin, Haining Zheng· July 31, 2026 View original

Key takeaways

  • A hybrid AI method quantifies exposed skin from images for dermal exposure assessment.
  • It combines Mask R-CNN for subject identification with a color-based segmentation algorithm.
  • The method achieved 80% agreement with human estimates on indoor painting images.
  • This offers a scalable and efficient tool for enhancing occupational safety monitoring.

Who benefits

Occupational SafetyConstructionManufacturingHealthcareChemicals

Summary

A new hybrid deep learning method combines Mask R-CNN with a color-based algorithm to quantify exposed skin from images, enabling enhanced dermal exposure assessment. The approach demonstrated approximately 80% agreement with human estimates using indoor painting images.

Researchers have developed a novel hybrid computer vision method designed to quantify the amount of exposed skin visible in images, a crucial step for assessing dermal exposure in various occupational settings. The technique integrates Mask R-CNN, a powerful instance segmentation model, with a specialized color-based algorithm. The process begins with Mask R-CNN identifying human subjects within an image and effectively removing background clutter. Subsequently, the color-based algorithm precisely segments the exposed skin regions. This two-stage approach was validated using a dataset of 170 indoor-painting images. Results showed that the pixel ratios of exposed skin to total body area derived by the system achieved approximately 80% agreement with estimates made by human observers. This demonstrates a scalable and semi-quantitative method for extracting valuable exposure information from visual data, with potential for future enhancements like body-part recognition and personal protective equipment (PPE) detection.

Why it matters

Accurate assessment of dermal exposure is vital for occupational safety and health. This AI-driven method offers a scalable, objective, and efficient way to monitor exposure risks from images, potentially reducing manual effort and improving safety protocols in various industries.

How to implement this in your domain

  1. 1Pilot this hybrid computer vision method for automated dermal exposure assessment in high-risk work environments.
  2. 2Integrate the system into existing safety monitoring workflows to provide semi-quantitative exposure data.
  3. 3Develop a dataset of images specific to your industry to train and fine-tune the model for optimal accuracy.
  4. 4Explore extending the system to detect specific body parts and identify the presence or absence of PPE.
  5. 5Utilize video streams for continuous, real-time exposure monitoring in dynamic work settings.

Original post by Hua Qian, Manisha Kotha, Tuan Tran, Jennifer Shin, Haining Zheng

"arXiv:2607.26170v1 Announce Type: cross Abstract: This study developed a hybrid computer vision method to quantify exposed skin from images for dermal exposure assessment. Using 170 indoor-painting images, Mask R-CNN first identified human subjects and removed background interfer…"

View on X

Originally posted by Hua Qian, Manisha Kotha, Tuan Tran, Jennifer Shin, Haining Zheng on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses

More in AI Research

AI ResearchAI Engineering & DevTools

New Framework Improves Partial Multi-View Clustering Performance.

DAS-PMVC is a novel framework for partial multi-view clustering that addresses view asymmetry and irrelevant samples by leveraging dual alignment and structure enhancement. It uses anchor graph structure alignment, structure-enhanced feature learning, and a dual alignment strategy to achieve superior clustering performance on various datasets.

Shubin Ma, Liang Zhao, Chuanye He, Zhenjiao Liu, Liang Zou, Lin Yuanbo Wu, Yu ShaoJul 31, 2026
AI Engineering & DevToolsAI Research

Dual Teachers Improve Adversarial Robustness and Accuracy.

This work extends Information Bottleneck Distillation (IBD) by introducing a "clean teacher" alongside a robust teacher to improve the robustness/accuracy tradeoff against adversarial attacks. The proposed method transfers features from both teachers to a student model, achieving better clean accuracy while maintaining adversarial robustness, outperforming original IBD and competing with state-of-the-art approaches.

Vincent Ryusuke Takahashi, Yoshinari Takeishi, Jun'ichi Takeuchi, Kave SalamatianJul 31, 2026
AI Engineering & DevToolsAI Research

Dynamic Batch Sizes Improve Large Language Model Training Efficiency.

This paper proposes a new approach to deep learning dynamics, deriving joint scaling laws for loss based on both learning rate and batch size schedules. It introduces an optimal dynamic batch size schedule that consistently outperforms static batch size baselines, highlighting its importance for large language model training.

Jiaxiang Li, Zhiqi Bu, Shiyun XuJul 31, 2026