Harnessing LLMs for Document-Guided Fuzzing of OpenCV Library
This program is tentative and subject to change.
The combination of computer vision and artificial intelligence is fundamentally transforming a broad spectrum of industries by enabling machines to interpret and act upon visual data with high-levels of accuracy. As the biggest and by far the most popular open-source computer vision library, OpenCV library provides an extensive suite of programming functions supporting real-time computer vision. Bugs in the OpenCV library can affect the downstream computer vision applications, and it is critical to ensure the reliability of the OpenCV library. This paper introduces VistaFuzz, a novel technique for harnessing large language models (LLMs) for document-guided fuzzing of the OpenCV library. It utilizes LLMs to standardize information from API documentation, extracting constraints on individual input parameters and dependencies between them based on this standardized API information to generate new input values for testing each target API. We evaluate the effectiveness of VistaFuzz in testing 330 APIs in the OpenCV library, and the results show that VistaFuzz detected 17 new bugs, where 10 bugs have been confirmed, and 5 of these have been fixed.
This program is tentative and subject to change.
Wed 10 SepDisplayed time zone: Auckland, Wellington change
13:30 - 15:00 | Session 4 - Testing 1Research Papers Track / Registered Reports / Journal First Track / NIER Track / Industry Track / Tool Demonstration Track at Case Room 2 260-057 Chair(s): Sigrid Eldh Ericsson AB, Mälardalen University, Carleton University | ||
13:30 15m | Performance Testing in Open-Source Web Projects: Adoption, Maintenance, and a Change Taxonomy Research Papers Track Sergio Di Meglio Università degli Studi di Napoli Federico II, Luigi Libero Lucio Starace Università degli Studi di Napoli Federico II, Valeria Pontillo Gran Sasso Science Institute, Ruben Opdebeeck Vrije Universiteit Brussel, Coen De Roover Vrije Universiteit Brussel, Sergio Di Martino Università degli Studi di Napoli Federico II Pre-print | ||
13:45 15m | Harnessing LLMs for Document-Guided Fuzzing of OpenCV Library Research Papers Track Bin Duan The University of Queensland, Tarek Mahmud Texas State University, Meiru Che Central Queensland University, Yan Yan University of Illinois Chicago, Naipeng Dong The University of Queensland, Australia, Dan Dongseong Kim The University of Queensland, Guowei Yang University of Queensland | ||
14:00 10m | XTestGen: Natural Language to Maintainable E2E Test Scripts with LLMs Tool Demonstration Track File Attached | ||
14:10 10m | Towards Effective Lightweight Test Oracles for Automated Multi-Fault Program Repair NIER Track Omar I. Al-Bataineh Gran Sasso Science Institute (GSSI) | ||
14:20 15m | Testing Is Not Boring: Characterizing Challenge in Software Testing Tasks Industry Track Davi Gama Hardman CESAR - Recife Center for Advanced Studies and Systems, César França Federal Rural University of Pernambuco (UFRPE), Brody Stuart-Verner University of Calgary, Ronnie de Souza Santos University of Calgary | ||
14:35 15m | Enriching automatic test case generation by extracting relevant test inputs from bug reports Journal First Track Wendkuuni Arzouma Marc Christian OUEDRAOGO University of Luxembourg, Laura Plein CISPA Helmholtz Center for Information Security, Abdoul Kader Kaboré University of Luxembourg, Andrew Habib ABB Corporate Research, Germany, Jacques Klein University of Luxembourg, David Lo Singapore Management University, Tegawendé F. Bissyandé University of Luxembourg | ||
14:50 10m | An Empirical Study of Complexity, Heterogeneity, and Compliance of GitHub Actions Workflows Registered Reports Edward Abrokwah Department of Computer Science, Trent University, Peterborough, Canada, Taher A. Ghaleb Trent University Pre-print | ||