Heng Huang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Heng Huang has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Xin Jin
Department of Cyber Security, Beijing Electronic Science and Technology Institute, Beijing, China
5 shared publications - Qianqian Qiao
Nanjing University, Nanjing, China
3 shared publications - Chaoen Xiao
Beijing Electronic Science and Technology Institute, Beijing, China
2 shared publications - Hao Lou
Beijing Electronic Science and Technology Institute, Beijing, China
2 shared publications - Xinning Li
Beijing Electronic Science and Technology of Institute, Beijing, China
2 shared publications - Aimin Hao
Beihang University, Beijing, China
1 shared publication
Latest Publications
- Open Access
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 8849-8857
VAC-Bench: From Text Similarity to Norm Conformance in Video Aesthetic Captioning Evaluation
Although Multimodal Large Language Models (MLLMs) have advanced rapidly in video understanding, their ability to produce video aesthetic captions remains poorly evaluated: existing metrics (e.g., BLEU, ROUGE) penalize linguistic diversity and cannot verify whether captions conform to established aesthetic norms. To bridge this gap, we present VAC-Bench, the first benchmark for systematic video aesthetic captioning evaluation, comprising 1,800 expert-annotated videos across diverse visual domains with fine-grained aesthetic attribute tags, quality ratings, and hierarchical captioning tasks. We further propose VANE (Video Aesthetic Norm Evaluation), a protocol that shifts assessment from reference text matching to norm conformance verification, introducing a three-tier classification that disentangles factual hallucinations from normative deviations and yielding three metrics: Norm Conformance, Coverage, and Completeness. Evaluation of 11 state-of-the-art MLLMs reveals a striking perception-expression gap: high dimensional coverage but substantially lower norm conformance, indicating that current models perceive which aesthetic dimensions to address yet struggle to articulate norm-conformant judgments. The code are available at https://github.com/flyingbird1993/VAC-Bench.
- 9 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 8849-8857
Open AccessVAC-Bench: From Text Similarity to Norm Conformance in Video Aesthetic Captioning Evaluation
Although Multimodal Large Language Models (MLLMs) have advanced rapidly in video understanding, their ability to produce video aesthetic captions remains poorly evaluated: existing metrics (e.g., BLEU, ROUGE) penalize linguistic diversity and cannot verify whether captions conform to established aesthetic norms. To bridge this gap, we present VAC-Bench, the first benchmark for systematic video aesthetic captioning evaluation, comprising 1,800 expert-annotated videos across diverse visual domains with fine-grained aesthetic attribute tags, quality ratings, and hierarchical captioning tasks. We further propose VANE (Video Aesthetic Norm Evaluation), a protocol that shifts assessment from reference text matching to norm conformance verification, introducing a three-tier classification that disentangles factual hallucinations from normative deviations and yielding three metrics: Norm Conformance, Coverage, and Completeness. Evaluation of 11 state-of-the-art MLLMs reveals a striking perception-expression gap: high dimensional coverage but substantially lower norm conformance, indicating that current models perceive which aesthetic dimensions to address yet struggle to articulate norm-conformant judgments. The code are available at https://github.com/flyingbird1993/VAC-Bench.
- Open Access
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 135-141
Multimodal Large Models for Image Tampering Detection and Explanation: From Detection to Reasoning
- 27 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 135-141
Open AccessMultimodal Large Models for Image Tampering Detection and Explanation: From Detection to Reasoning
- Open Access
- 26 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 27-35
Optimizing Similar Image Selection through Multimodal Feature Fusion
- 26 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 27-35
Open AccessOptimizing Similar Image Selection through Multimodal Feature Fusion
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 63-71
Predicting Scores of Various Aesthetic Attribute Sets by Learning from Overall Score Labels
For aesthetic attribute evaluation of images (AAEI), the annotation of image aesthetic attribute scores plays an important role. It requires experienced artists and professional photographers, which hinders the collection of large-scale fully-annotated datasets. In this paper, we propose to learn an AAEI model from the datasets with only the overall aesthetic scores of the whole image. First, a novel aesthetic attribute evaluation framework based on attribute features is proposed to predict attribute scores and overall scores. We call it the F2S (attribute features to attribute scores) model. Our F2S model is trained from instances with only overall aesthetic scores, bridging the gap between overall and attribute scores. Second, we define an aesthetic attribute contribution to describe the role of aesthetic attributes throughout an image and use the attention weights from a channel attention module as contribution values. Last, we propose the relationship among the attribute scores, the attribute contributions, and the overall scores and use the relationship to train our F2S model. Sufficient experiments on publicly available datasets demonstrate that our F2S model achieves comparable performance with those trained on the datasets with fully annotated aesthetic attribute score labels. Using our method, it is feasible to learn meaningful attribute scores for various aesthetic attribute sets of different types of images with only overall aesthetic scores, such as subject and composition for landscapes and emotion and pose for portraits.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 63-71
Open AccessPredicting Scores of Various Aesthetic Attribute Sets by Learning from Overall Score Labels
For aesthetic attribute evaluation of images (AAEI), the annotation of image aesthetic attribute scores plays an important role. It requires experienced artists and professional photographers, which hinders the collection of large-scale fully-annotated datasets. In this paper, we propose to learn an AAEI model from the datasets with only the overall aesthetic scores of the whole image. First, a novel aesthetic attribute evaluation framework based on attribute features is proposed to predict attribute scores and overall scores. We call it the F2S (attribute features to attribute scores) model. Our F2S model is trained from instances with only overall aesthetic scores, bridging the gap between overall and attribute scores. Second, we define an aesthetic attribute contribution to describe the role of aesthetic attributes throughout an image and use the attention weights from a channel attention module as contribution values. Last, we propose the relationship among the attribute scores, the attribute contributions, and the overall scores and use the relationship to train our F2S model. Sufficient experiments on publicly available datasets demonstrate that our F2S model achieves comparable performance with those trained on the datasets with fully annotated aesthetic attribute score labels. Using our method, it is feasible to learn meaningful attribute scores for various aesthetic attribute sets of different types of images with only overall aesthetic scores, such as subject and composition for landscapes and emotion and pose for portraits.
- Open Access
- 29 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 127-135
Language Guidance Generation Using Aesthetic Attribute Comparison for Human Photography and AIGC
With the proliferation of mobile photography technology, leading mobile phone manufacturers are racing to enhance the shooting capabilities of their equipment and the photo beautification algorithm of their software. However, the development of intelligent equipment and algorithms cannot supplant human subjective photography techniques. Simultaneously, with the rapid advancement of AIGC technology, AI simulation shooting has become an integral part of people's daily lives. If it were possible to assist human photography and AIGC with language guidance, this would be a significant step forward in subjectively improving the aesthetic quality of photographic images. In this paper, we propose Aesthetic Language Guidance of Image (ALG) and present a series of language guidance rules (ALG Rules). ALG is divided into ALG-T and ALG-I based on whether the guiding rules are derived from photography templates or reference images, respectively. ALG-T and ALG-I both provide guidance for photography based on three attributes of color, light, and composition of images. ALG-T and ALG-I provide aesthetic language guidance for two types of input images, landscape and portrait images. We employ two methods to conduct confirmatory experiments, human photography, and AIGC imitation shooting. In the experiments, by comparing the aesthetic scores of original and modified images, the results show that our proposed guidance scheme significantly improves the aesthetic quality of photos in terms of color, composition, and lighting attributes.
- 29 October 2023
- Published by Association for Computing Machinery (ACM)
- p. 127-135
Open AccessLanguage Guidance Generation Using Aesthetic Attribute Comparison for Human Photography and AIGC
With the proliferation of mobile photography technology, leading mobile phone manufacturers are racing to enhance the shooting capabilities of their equipment and the photo beautification algorithm of their software. However, the development of intelligent equipment and algorithms cannot supplant human subjective photography techniques. Simultaneously, with the rapid advancement of AIGC technology, AI simulation shooting has become an integral part of people's daily lives. If it were possible to assist human photography and AIGC with language guidance, this would be a significant step forward in subjectively improving the aesthetic quality of photographic images. In this paper, we propose Aesthetic Language Guidance of Image (ALG) and present a series of language guidance rules (ALG Rules). ALG is divided into ALG-T and ALG-I based on whether the guiding rules are derived from photography templates or reference images, respectively. ALG-T and ALG-I both provide guidance for photography based on three attributes of color, light, and composition of images. ALG-T and ALG-I provide aesthetic language guidance for two types of input images, landscape and portrait images. We employ two methods to conduct confirmatory experiments, human photography, and AIGC imitation shooting. In the experiments, by comparing the aesthetic scores of original and modified images, the results show that our proposed guidance scheme significantly improves the aesthetic quality of photos in terms of color, composition, and lighting attributes.
- 5 September 2022
- Published by Elsevier in Computers and Electrical Engineering
- Vol. 103
Aesthetic evaluation of Asian and Caucasian photos with overall and attribute scores
- 5 September 2022
- Published by Elsevier in Computers and Electrical Engineering
- Vol. 103
Aesthetic evaluation of Asian and Caucasian photos with overall and attribute scores