# A Deep Learning Approach to Interface Color Quality Assessment in HCI

arXiv: https://arxiv.org/abs/2502.09914 · HTML: https://mlsanity.com/paper/2502.09914/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Feb 2025.

## Built on by (most cited first, 5 of 5)

- [Deep Probabilistic Modeling of User Behavior for Anomaly Detection via Mixture Density Networks](https://mlsanity.com/paper/2505.08220/) (2025)
- [A LongFormer-Based Framework for Accurate and Efficient Medical Text Summarization](https://mlsanity.com/paper/2503.06888/) (2025)
- [Semantic and Contextual Modeling for Malicious Comment Detection with BERT-BiLSTM](https://mlsanity.com/paper/2503.11084/) (2025)
- [Intelligent Task Scheduling for Microservices via A3C-Based Reinforcement Learning](https://mlsanity.com/paper/2505.00299/) (2025)
- [Enhanced Web User Interface Design Via Cross-Device Responsiveness Assessment Using An Improved HCI-INTEGRATED DL Schemes](https://mlsanity.com/paper/2512.15775/) (2025)

## Abstract (arXiv)

In this paper, a quantitative evaluation model for the color quality of human-computer interaction interfaces is proposed by combining deep convolutional neural networks (CNN). By extracting multidimensional features of interface images, including hue, brightness, purity, etc., CNN is used for efficient feature modeling and quantitative analysis, and the relationship between interface design and user perception is studied. The experiment is based on multiple international mainstream website interface datasets, covering e-commerce platforms, social media, education platforms, etc., and verifies the evaluation effect of the model on indicators such as contrast, clarity, color coordination, and visual appeal. The results show that the CNN evaluation is highly consistent with the user rating, with a correlation coefficient of up to 0.96, and it also shows high accuracy in mean square error and absolute error. Compared with traditional experience-based evaluation methods, the proposed model can efficiently and scientifically capture the visual characteristics of the interface and avoid the influence of subjective factors. Future research can explore the introduction of multimodal data (such as text and interactive behavior) into the model to further enhance the evaluation ability of dynamic interfaces and expand it to fields such as smart homes, medical systems, and virtual reality. This paper provides new methods and new ideas for the scientific evaluation and optimization of interface design.