Qiang Li (利 强)

From Understanding to Engagement: Personalized Pharmacy Video Clips via Vision Language Models (VLMs)

Suyash Mishra, Qiang Li, Anubhav Girdhar, Srikanth Patil

ACL 2026 · Workshop on Customized and Personalized NLP (CustomNLP4U) · Poster Paper

A domain-adapted video-to-video clip generation framework combining Audio- and Vision-Language Models to produce highlight clips: a reproducible Cut & Merge algorithm with fade transitions and timestamp normalization, a role-based personalization mechanism for marketing/training/regulatory outputs, and a cost-efficient end-to-end pipeline. Evaluated on Video-MME and a proprietary set of 16,159 pharmacy videos across 14 disease areas, showing 3–4× speedup and 4× cost reduction over VLM baselines such as Gemini 2.5 Pro.

Scaling Vision–Language Models for Pharmaceutical Long-Form Video Reasoning on Industrial GenAI Platform

Suyash Mishra, Qiang Li, Srikanth Patil, Satyanarayan Pati, Baddu Narendra

ACL 2026 · Workshop on Advances in Language and Vision Research (ALVR) · 🌟 Spotlight Paper

An industrial GenAI framework processing 200,000+ PDFs, 25,326 videos across eight formats, and 888 multilingual audio files in 20+ languages. Contributions: a large-scale architecture for multimodal reasoning in pharma; empirical analysis of 40+ VLMs on Video-MME, MMBench, and a proprietary 25,326-video set across 14 disease areas; and findings on multimodality, attention trade-offs, temporal reasoning limits, and video-splitting under GPU constraints.

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

NAACL 2025 · Industry Track

A cross-lingual benchmark of nearly 99,869 real traffic incident records from Vienna (2013–2023) assessing LLM robustness across spatial and temporal domains. Explores three hypotheses — sentence indexing, date-to-text conversion, and German-to-English translation — and incorporates RAG to further examine hallucination in both domains.

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

CVPR 2024 · Synthetic Data for Computer Vision Workshop · Full paper

An explainable visual dataset of 200K+ images with 15,410 manual semantic annotations across 12 ImageNet categories representing everyday objects. Incorporates six real-world scenarios (overexposure, blurring, color shifts, etc.) and a quantitative robustness criterion, with particular attention to background influence.

Self-Supervised Learning with Temporary Exact Solutions: Linear Projection

Evrim Ozmermer, Qiang Li

INDIN 2023 · IEEE 21st Int'l Conference on Industrial Informatics · Full paper

A self-supervised training method for visual transformers that learns meaningful image/video representations without large labeled sets, based on exact solutions of generated representations. The learned features fine-tune effectively on industrial downstream tasks.

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

ICASSP 2023 · Full paper

SleepHGNN, a deep model for sleep-stage classification using a Heterogeneous Graph Transformer to capture interactivity and heterogeneity across multimodal signals — the first attempt to apply heterogeneous graph neural networks to this task, with a graph-level classification framework generalizable to domains like protein and molecular graph classification.

Continual Learning on Deployment Pipelines for Machine Learning Systems

Qiang Li, Chongyu Zhang

NeurIPS 2022 · DMML Workshop

A comparison of ML deployment solutions spanning manual training/deployment through automated continuous-integration workflows, with practical evaluation metrics and a look at how real-world requirements diverge from academic settings.

Explainable AI: Object Recognition With Help From Background

Raza Hashmi, Qiang Li

ICLR 2022 · CSS Workshop

An exploration of how image backgrounds can help object recognition, building on the "noise or signal" baseline work by Xiao et al.

AI Quality Next — BMW Group Computer Vision Project

Computer Vision Engineer: Qiang Li

BMW Group · Industrial deployment

AIQX integrates machine learning and deep learning algorithms for visual inspection directly into BMW's production processes — a central standard for AI-based quality inspection across the global production system, enabling more robust defect detection and order verification.

All You Need Is Cell Attention: A Cell Annotation Tool for Single-Cell Morphology Data

Qiang Li*, Corin Otesteanu, Lily Xu

ICLR 2021 · Workshop on AI for Public Health

Cell Morphology Based Diagnosis of Cancer using Convolutional Neural Networks: CellNet

Qiang Li, Yiran Xing, Tianwei Lan, ChenYu Tian, Ying Chen

DeeCamp 2020 · Medical Track of AI in Public Healthcare — Challenge winner

Localization and Visualization of Defects by PCA, KMeans, Colorspace Template Matching for Additive Manufacturing

Hamid Jahangir, Qiang Li

ICAM 2020 · International Conference on Additive Manufacturing — Invited talk

GPT-3 Industry Survey and Applied Scenarios

Qiyi Ye, Qiang Li

Sinovation Ventures AI Institute (创新工场), 2020

Designed three GPT-based generative models for real-world business scenarios.