M.Sc. Informatik

  RWTH Aachen University, Aachen, Germany

  Email: qiang.li@rwth-aachen.de

Guten Tag! I am currently working as an Associate Manager at Accenture. Prior to joining Accenture, I served as an IDEA Research Grant Student in Prof. Dr. Manfred Claassen group at ETH Zürich, and I obtained my Master's degree in Informatik from RWTH Aachen University, bachelor's degree from HFUT. During my Bachelor's studies, I specialized in IoT and founded the HFUT Robocup Lab. I also gained industry experience as a computer vision working student at Siemens AG Aachen Gas Turbine Research Center during my Master’s studies.

My research interests primarily focus on object recognition and segmentation, the deployment of machine learning systems, model interpretability, multi-task and multimodal learning in industrial production.

[Feedback from Mentor at ETH Claassen Lab]

[Feedback from Mentor at Sinovation Venture AI Institute]

[Feedback from Mentor at RWTH Computer Vision Group]

From Understanding to Engagement: Personalized pharmacy Video Clips via Vision Language Models (VLMs)
Suyash Mishra, Qiang Li, Anubhav Girdhar, Srikanth Patil
Accepted by The 64th Annual Meeting of the Association for Computational Linguistics (ACL'26), ACL 2026 Workshop CustomNLP Poster Paper , 2026.
In this work, we introduce a domain-adapted Video-to-Video Clip Generation framework that integrates Audio-Language Models (ALMs) and Vision Language Models (VLMs) to produce highlight clips. Our contributions are threefold: (i) a reproducible Cut & Merge algorithm with fade-in/out and timestamp normalization, ensuring smooth transitions and audio/visual alignment; (ii) a personalization mechanism based on role definition and prompt injection for tailored outputs (marketing, training, regulatory); (iii) a cost-efficient e2e pipeline strategy balancing ALM/VLM-enhanced processing. Evaluations on Video-MME benchmark (900) and our proprietary dataset of 16,159 pharmacy videos across 14 disease areas demonstrate 3–4× speedup, 4× cost reduction, and competitive clip quality. Beyond efficiency gains, we also report our methods improved clip coherence scores (0.348) and informativeness scores (0.721) over state-of-the-art VLM baselines (e.g., Gemini 2.5 Pro). Paper    Conference Website    Arxiv   

Scaling Vision–Language Models for Pharmaceutical Long-Form Video Reasoning on Industrial GenAI Platform
Suyash Mishra, Qiang Li, Srikanth Patil, Satyanarayan Pati, Baddu Narendra
Accepted by The 64th Annual Meeting of the Association for Computational Linguistics (ACL'26), Workshop on Advances in Language and Vision Research (ALVR), 🌟 Spotlight Paper , 2026.
In this work, we present an industrial GenAI framework that processes over 200,000 PDFs, 25,326 videos across eight formats (e.g., MP4, M4V, etc.), and 888 multilingual audio files in more than 20 languages. Our study makes three contributions: (i) an industrial large-scale architecture for multimodal reasoning in pharmaceutical domains; (ii) empirical analysis of over 40 VLMs on two leading benchmarks (Video-MME and MMBench) and proprietary dataset of 25,326 videos across 14 disease areas; and (iii) four findings relevant to long-form video reasoning: the role of multimodality, attention mechanism trade-offs, temporal reasoning limits, and challenges of video splitting under GPU constraints. Paper    Conference Website    Accept Spotlight Lists (1/6)    

How LLMs React to Industrial Spatio-Temporal Data? Assessing Hallucination with a Novel Traffic Incident Benchmark Dataset
Qiang Li, Mingkun Tan, Xun Zhao, Dan Zhang, Daoan Zhang, Shengzhao Lei, Anderson S Chu, Lujun Li, Porawit Kamnoedboon
Accepted by 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL'25), Full paper, 2025.
This study, originating from a real world industrial GenAI application, introduces a novel cross-lingual benchmark dataset comprising nearly 99,869 real traffic incident records from Vienna (2013-2023) to assess the robustness of state-of-the-art LLMs in the spatial and temporal domains for traffic incident classification. We then explored three hypotheses — sentence indexing, date-to-text conversion, and German-to-English translation — and incorporated Retrieval Augmented Generation (RAG) to further examine LLM hallucinations in both spatial and temporal domains. Paper    Code    Conference Website    Dataset    

XIMAGENET-12: An Explainable Visual Benchmark Dataset for Model Robustness Evaluation
Qiang Li, Dan Zhang, Shengzhao Lei, Xun Zhao, Porawit Kamnoedboon, WeiWei Li, Junhao Dong, Shuyan Li
Accepted by Synthetic Data for Computer Vision Workshop @ IEEE (CVPR'24), Full paper, 2024.
In this paper, we propose an explainable visual dataset, XIMAGENET-12, to evaluate the robustness of visual models. XIMAGENET-12 consists of over 200K images with 15,410 manual semantic annotations. Specifically, we deliberately selected 12 categories from ImageNet, representing objects commonly encountered in practical life. To simulate real-world situations, we incorporated six diverse scenarios, such as overexposure, blurring, and color changes, etc. We further develop a quantitative criterion for robustness assessment, allowing for a nuanced understanding of how visual models perform under varying conditions, notably in relation to the background.
Paper    Code    Conference Website    Dataset Download (126+ Download!)    

Self-supervised Learning with Temporary Exact Solutions: Linear Projection
Evrim Ozmermer, Qiang Li
Accepted by IEEE 21st International Conference on Industrial Informatics (INDIN'23), Full paper, 2023.
In this paper, we present a self-supervised learning method for training, not limited to but especially visual transformers that are able to learn meaningful representations of images and videos without requiring large amounts of labeled data. Our method is based on using exact solutions of the representations that the model generates. It is shown that the model is able to learn useful features that can be later fine-tuned on industrial downstream tasks.
Paper    Code    Conference Website   

Exploiting Interactivity and Heterogeneity for Sleep Stage Classification via Heterogeneous Graph Neural Network
Ziyu Jia, Youfang Lin, Yuhan Zhou, Xiyang Cai, Peng Zheng, Qiang Li, Jing Wang
Accepted by 2023 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP),Full paper, 2023.
In this paper, we propose a novel deep model SleepHGNN for sleep stage classification. The Heterogeneous Graph Transformer is applied to capture the interactivity and heterogeneity of the multimodal signals. To the best of our knowledge, this is the first attempt to leverage heterogeneous graph neural networks for sleep stage classification. Since the SleepHGNN is a universal framework for the graph-level classification task based on the heterogeneous graph, we will generalize model to other domains like protein classification and molecular graph classification including anticancer chemical compound classification in the future.
Paper    Code    Conference Website   

Continual learning on deployment pipelines for Machine Learning Systems
Qiang Li, Chongyu Zhang
Accepted by The Conference on Neural Information Processing Systems (NeurIPS), DMML Workshop, 2022.
In this work, we presented a comparison of various solutions for the deployment of machine learning systems, includes different layers of automation from highly manual model training and deployment to an automated continuous integration workflow. We proposed the evaluation metrics in practice and described how real-world requirements differ from more academic settings. Livestream on NeurIPS2022    Poster    Paper on arxiv    ResearchGate    Workshop Website   

Explainable AI: Object Recognition With Help From Background
Raza Hashmi, Qiang Li
Accepted by The International Conference on Learning Representations (ICLR), CSS Workshop, 2022.
This work explores how backgrounds might help in object recognition tasks in depth. Our project is fascinated by the baseline work done by Xiao et al. in their noise or signal paper. Website    Blogs    Camera ready video of presentation on ICLR2022    Dataset on Kaggle (184+ Download!)   

AI Quality Next - BMW Group - Computer Vision project
Computer Vision Engineer: Qiang Li
AIQX provides a platform to integrate machine learning & deep learning algorithms for visual inspections directly into the production processes. Creation of a central standard for the implementation of AI for quality inspections in the global production system. AI provides far more robust algorithms and opens new areas of defect detection and order verification for manufacturing.
Youtube:Artificial Intelligence at the BMW Group    Youtube:BMW Factory – Integration of A.I. in the Production Line    Published materials about our AIQX Project & Usecases in News and Major Media! Won BMW Q-Award*    Our AIQX Project & Usecases in CNBC News*   

All You Need Is Cell Attention: A Cell Annotation Tool for Single-Cell Morphology Data
Qiang Li*, Corin Otesteanu, Lily Xu
Accepted by The International Conference on Learning Representations (ICLR), Workshop on AI for Public Health, 2021.
PDF    Code/Software    

Cell Morphology Based Diagnosis of Cancer using Convolutional Neural Networks: CellNet
Qiang Li, Yiran Xing, Tianwei Lan, ChenYu Tian, Ying Chen
Won Challenge on Medical Track of AI in Public Healthcare of DeeCamp 2020.
Website    Code/Models   

Localization and visualization of defects by PCA, KMeans, Colorspace Template Matching for Additive Manufacturing
Hamid Jahangir, Qiang Li
Invited Talk on International Conference on Additive Manufacturing (ICAM), 2020.
Slides    Software   

GPT-3 industry survey and applied scenarios
Qiyi Ye, Qiang Li
Designed 3 GPT-based generative model on real scenario at Sinovation Ventures AI Institute （创新工场）, 2020.
Code/GPT generative models    Slides

Hobbies: Vloger who loves Museum, Cooking... You can find me in TikTok / Red(小红书)/ Wechat Channel by ‘Jonas的新鲜感’（Jonas' curiosity), We have received millions of views and likes 👍, and thousands of followers! Have created 100+ Vlogs. Keeping Learning! Let's move on together!
I am a fan of Hackathons. It gave me valuable experiences and developed critical thinking, problem-solving, and leadership skills.