Dinesh Babu Jayagopi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Dinesh Babu Jayagopi has a total of 139 co-authors. Here are the top co-authors by number of shared publications:
Daniel Gatica-Perez
Idiap-EPFL, Switzerland
14 shared publicationsJean-Marc Odobez
IDIAP Research Institute, Martigny, Switzerland
6 shared publicationsSowmya Rasipuram
International Institute of Information Technology, Bangalore, Karnataka, India
6 shared publications- Chinchu Thomas
Multimodal Perception Lab, International Institute of Information Technology (IIITB), Bangalore, India
5 shared publications - Mauro Cherubini
University of Lausanne
5 shared publications Hayley Hung
IDIAP Research Institute, Martigny, Switzerland
4 shared publications
Latest Publications
- Open Access
- 4 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 883-891
HAAS: Holistic Attention-free Animation from Speech using Mamba
- 4 October 2026
- Published by Association for Computing Machinery (ACM)
- p. 883-891
Open AccessHAAS: Holistic Attention-free Animation from Speech using Mamba
- Open Access
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (20) , 47901-47914
OMuSense-23: A Multimodal Dataset for Contactless Breathing Pattern Recognition and Biometric Analysis
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (20) , 47901-47914
Open AccessOMuSense-23: A Multimodal Dataset for Contactless Breathing Pattern Recognition and Biometric Analysis
- Open Access
- 25 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 8202-8222
“Robotic answers”: Participatory Identification of Risks in AI-Assisted Interview Training by Job Seekers
- 25 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 8202-8222
Open Access“Robotic answers”: Participatory Identification of Risks in AI-Assisted Interview Training by Job Seekers
- 9 June 2026
- Published by Elsevier in Infant Behavior and Development
- Vol. 84, 102205
Multimodal behavioral analysis of child play interactions for early detection of Autism Spectrum Disorder
- 9 June 2026
- Published by Elsevier in Infant Behavior and Development
- Vol. 84, 102205
Multimodal behavioral analysis of child play interactions for early detection of Autism Spectrum Disorder
- Open Access
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Early Diagnosis of Autism through Behaviors from Semi-Structured Triadic Play Interaction Videos
Computational analysis of children’s play-based behaviors provides insights into subtle atypical traits associated with Autism Spectrum Disorder (ASD). Accurately characterizing these behaviors is crucial for building predictive models that support early ASD diagnosis. We propose a two-stage video-based pipeline for ASD detection. In the first stage, the behavioral detection model identifies elicited behaviors during semi-structured triadic play sessions. This stage leverages features extracted from OpenFace, OpenPose, and Swin-2D representations of both objects and individuals involved in the interaction. The second stage, ASD classification, uses the detected behaviors to predict the likelihood of an ASD diagnosis. The framework leverages Conv1D, ResNet, and multi-scale ConvTransformers to model both local and global feature relationships, effectively capturing short- and long-term temporal dynamics for behavior detection and ASD prediction. Experimental results demonstrate the effectiveness of the proposed pipeline, achieving a maximum mean Average Precision (mAP) of 46 across 18 behavior classes in the behavioral detection module and a maximum F1-score of 0.80 for the ASD classification module. Finally, an integrated explainability component provides insights into the significance of various behavioral patterns across video clips.
- 17 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-10
Open AccessEarly Diagnosis of Autism through Behaviors from Semi-Structured Triadic Play Interaction Videos
Computational analysis of children’s play-based behaviors provides insights into subtle atypical traits associated with Autism Spectrum Disorder (ASD). Accurately characterizing these behaviors is crucial for building predictive models that support early ASD diagnosis. We propose a two-stage video-based pipeline for ASD detection. In the first stage, the behavioral detection model identifies elicited behaviors during semi-structured triadic play sessions. This stage leverages features extracted from OpenFace, OpenPose, and Swin-2D representations of both objects and individuals involved in the interaction. The second stage, ASD classification, uses the detected behaviors to predict the likelihood of an ASD diagnosis. The framework leverages Conv1D, ResNet, and multi-scale ConvTransformers to model both local and global feature relationships, effectively capturing short- and long-term temporal dynamics for behavior detection and ASD prediction. Experimental results demonstrate the effectiveness of the proposed pipeline, achieving a maximum mean Average Precision (mAP) of 46 across 18 behavior classes in the behavioral detection module and a maximum F1-score of 0.80 for the ASD classification module. Finally, an integrated explainability component provides insights into the significance of various behavioral patterns across video clips.
- 8 November 2025
- Published by Springer Nature
- p. 207-222
Shan Handwritten Alphabet Recognition System: A Comparative Study of Machine Learning and Deep Learning Method
- 8 November 2025
- Published by Springer Nature
- p. 207-222
Shan Handwritten Alphabet Recognition System: A Comparative Study of Machine Learning and Deep Learning Method
- Open Access
- 26 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 75-84
VLM-Guided Toddler Behavior Recognition from Semi-Structured Triadic Play Videos
- 26 October 2025
- Published by Association for Computing Machinery (ACM)
- p. 75-84
Open AccessVLM-Guided Toddler Behavior Recognition from Semi-Structured Triadic Play Videos
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2164-2176
Invisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models
Artificial Intelligence (AI) is increasingly used in hiring, with large language models (LLMs) having the potential to influence or even make hiring decisions. However, this raises pressing concerns about bias, fairness, and trust, particularly across diverse cultural contexts. Despite their growing role, few studies have systematically examined the potential biases in AI-driven hiring evaluation across cultures. In this study, we conduct a systematic analysis of how LLMs assess job interviews across cultural and identity dimensions. Using two datasets of interview transcripts, 100 from UK and 100 from Indian job seekers, we first examine cross-cultural differences in LLM-generated scores for hirability and related traits. Indian transcripts receive consistently lower scores than UK transcripts, even when they were anonymized, with disparities linked to linguistic features such as sentence complexity and lexical diversity. We then perform controlled identity substitutions (varying names by gender, caste, and region) within the Indian dataset to test for name-based bias. These substitutions do not yield statistically significant effects, indicating that names alone, when isolated from other contextual signals, may not influence LLM evaluations. Our findings underscore the importance of evaluating both linguistic and social dimensions in LLM-driven evaluations and highlight the need for culturally sensitive design and accountability in AI-assisted hiring.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2164-2176
Open AccessInvisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models
Artificial Intelligence (AI) is increasingly used in hiring, with large language models (LLMs) having the potential to influence or even make hiring decisions. However, this raises pressing concerns about bias, fairness, and trust, particularly across diverse cultural contexts. Despite their growing role, few studies have systematically examined the potential biases in AI-driven hiring evaluation across cultures. In this study, we conduct a systematic analysis of how LLMs assess job interviews across cultural and identity dimensions. Using two datasets of interview transcripts, 100 from UK and 100 from Indian job seekers, we first examine cross-cultural differences in LLM-generated scores for hirability and related traits. Indian transcripts receive consistently lower scores than UK transcripts, even when they were anonymized, with disparities linked to linguistic features such as sentence complexity and lexical diversity. We then perform controlled identity substitutions (varying names by gender, caste, and region) within the Indian dataset to test for name-based bias. These substitutions do not yield statistically significant effects, indicating that names alone, when isolated from other contextual signals, may not influence LLM evaluations. Our findings underscore the importance of evaluating both linguistic and social dimensions in LLM-driven evaluations and highlight the need for culturally sensitive design and accountability in AI-assisted hiring.
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2150-2163
The AI Model Risk Catalog: What Developers and Researchers Miss About Real-World AI Harms
We analyzed nearly 460,000 AI model cards from Hugging Face to examine how developers report risks. From these, we extracted around 3,000 unique risk mentions and built the AI Model Risk Catalog. We compared these with risks identified by researchers in the MIT Risk Repository and with real-world incidents from the AI Incident Database. Developers focused on technical issues like bias and safety, while researchers emphasized broader social impacts. Both groups paid little attention to fraud and manipulation, which are common harms arising from how people interact with AI. Our findings show the need for clearer, structured risk reporting that helps developers think about human-interaction and systemic risks early in the design process.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2150-2163
Open AccessThe AI Model Risk Catalog: What Developers and Researchers Miss About Real-World AI Harms
We analyzed nearly 460,000 AI model cards from Hugging Face to examine how developers report risks. From these, we extracted around 3,000 unique risk mentions and built the AI Model Risk Catalog. We compared these with risks identified by researchers in the MIT Risk Repository and with real-world incidents from the AI Incident Database. Developers focused on technical issues like bias and safety, while researchers emphasized broader social impacts. Both groups paid little attention to fraud and manipulation, which are common harms arising from how people interact with AI. Our findings show the need for clearer, structured risk reporting that helps developers think about human-interaction and systemic risks early in the design process.
- Open Access
Invisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models
Artificial Intelligence (AI) is increasingly used in hiring, with large language models (LLMs) having the potential to influence or even make hiring decisions. However, this raises pressing concerns about bias, fairness, and trust, particularly across diverse cultural contexts. Despite their growing role, few studies have systematically examined the potential biases in AI-driven hiring evaluation across cultures. In this study, we conduct a systematic analysis of how LLMs assess job interviews across cultural and identity dimensions. Using two datasets of interview transcripts, 100 from UK and 100 from Indian job seekers, we first examine cross-cultural differences in LLM-generated scores for hirability and related traits. Indian transcripts receive consistently lower scores than UK transcripts, even when they were anonymized, with disparities linked to linguistic features such as sentence complexity and lexical diversity. We then perform controlled identity substitutions (varying names by gender, caste, and region) within the Indian dataset to test for name-based bias. These substitutions do not yield statistically significant effects, indicating that names alone, when isolated from other contextual signals, may not influence LLM evaluations. Our findings underscore the importance of evaluating both linguistic and social dimensions in LLM-driven evaluations and highlight the need for culturally sensitive design and accountability in AI-assisted hiring.
Open AccessInvisible Filters: Cultural Bias in Hiring Evaluations Using Large Language Models
Artificial Intelligence (AI) is increasingly used in hiring, with large language models (LLMs) having the potential to influence or even make hiring decisions. However, this raises pressing concerns about bias, fairness, and trust, particularly across diverse cultural contexts. Despite their growing role, few studies have systematically examined the potential biases in AI-driven hiring evaluation across cultures. In this study, we conduct a systematic analysis of how LLMs assess job interviews across cultural and identity dimensions. Using two datasets of interview transcripts, 100 from UK and 100 from Indian job seekers, we first examine cross-cultural differences in LLM-generated scores for hirability and related traits. Indian transcripts receive consistently lower scores than UK transcripts, even when they were anonymized, with disparities linked to linguistic features such as sentence complexity and lexical diversity. We then perform controlled identity substitutions (varying names by gender, caste, and region) within the Indian dataset to test for name-based bias. These substitutions do not yield statistically significant effects, indicating that names alone, when isolated from other contextual signals, may not influence LLM evaluations. Our findings underscore the importance of evaluating both linguistic and social dimensions in LLM-driven evaluations and highlight the need for culturally sensitive design and accountability in AI-assisted hiring.