_version_ 1866914124077727744
author Wang, Jiaqi
Yang, Xiao
Sun, Kai
Suresh, Parth
Sharma, Sanat
Czyzewski, Adam
Andersen, Derek
Appini, Surya
Banerjee, Arkav
Choudhary, Sajal
Ghasemlou, Shervin
Guan, Ziqiang
Iyer, Akil
Khan, Haidar
Kong, Lingkun
Luo, Roy
Ma, Tiffany
Qiao, Zhen
Tran, David
Xu, Wenfang
Yeatman, Skyler
Zhou, Chen
Gujral, Gunveer
Xia, Yinglong
Moon, Shane
Scheffer, Nicolas
Shah, Nirav
Chang, Eun
Liu, Yue
Metze, Florian
Stark, Tammy
Feizollahi, Zhaleh
Jessee, Andrea
Pujari, Mangesh
Aly, Ahmed
Damavandi, Babak
Wanga, Rakesh
Kumar, Anuj
Patel, Rohit
Yih, Wen-tau
Dong, Xin Luna
author_facet Wang, Jiaqi
Yang, Xiao
Sun, Kai
Suresh, Parth
Sharma, Sanat
Czyzewski, Adam
Andersen, Derek
Appini, Surya
Banerjee, Arkav
Choudhary, Sajal
Ghasemlou, Shervin
Guan, Ziqiang
Iyer, Akil
Khan, Haidar
Kong, Lingkun
Luo, Roy
Ma, Tiffany
Qiao, Zhen
Tran, David
Xu, Wenfang
Yeatman, Skyler
Zhou, Chen
Gujral, Gunveer
Xia, Yinglong
Moon, Shane
Scheffer, Nicolas
Shah, Nirav
Chang, Eun
Liu, Yue
Metze, Florian
Stark, Tammy
Feizollahi, Zhaleh
Jessee, Andrea
Pujari, Mangesh
Aly, Ahmed
Damavandi, Babak
Wanga, Rakesh
Kumar, Anuj
Patel, Rohit
Yih, Wen-tau
Dong, Xin Luna
contents Wearable devices such as smart glasses are transforming the way people interact with their surroundings, enabling users to seek information regarding entities in their view. Multi-Modal Retrieval-Augmented Generation (MM-RAG) plays a key role in supporting such questions, yet there is still no comprehensive benchmark for this task, especially regarding wearables scenarios. To fill this gap, we present CRAG-MM -- a Comprehensive RAG benchmark for Multi-modal Multi-turn conversations. CRAG-MM contains a diverse set of 6.5K (image, question, answer) triplets and 2K visual-based multi-turn conversations across 13 domains, including 6.2K egocentric images designed to mimic captures from wearable devices. We carefully constructed the questions to reflect real-world scenarios and challenges, including five types of image-quality issues, six question types, varying entity popularity, differing information dynamism, and different conversation turns. We design three tasks: single-source augmentation, multi-source augmentation, and multi-turn conversations -- each paired with an associated retrieval corpus and APIs for both image-KG retrieval and webpage retrieval. Our evaluation shows that straightforward RAG approaches achieve only 32% and 43% truthfulness on CRAG-MM single- and multi-turn QA, respectively, whereas state-of-the-art industry solutions have similar quality (32%/45%), underscoring ample room for improvement. The benchmark has hosted KDD Cup 2025, attracting about 1K participants and 5K submissions, with winning solutions improving baseline performance by 28%, highlighting its early impact on advancing the field.
format Preprint
id arxiv_https___arxiv_org_abs_2510_26160
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CRAG-MM: Multi-modal Multi-turn Comprehensive RAG Benchmark
Wang, Jiaqi
Yang, Xiao
Sun, Kai
Suresh, Parth
Sharma, Sanat
Czyzewski, Adam
Andersen, Derek
Appini, Surya
Banerjee, Arkav
Choudhary, Sajal
Ghasemlou, Shervin
Guan, Ziqiang
Iyer, Akil
Khan, Haidar
Kong, Lingkun
Luo, Roy
Ma, Tiffany
Qiao, Zhen
Tran, David
Xu, Wenfang
Yeatman, Skyler
Zhou, Chen
Gujral, Gunveer
Xia, Yinglong
Moon, Shane
Scheffer, Nicolas
Shah, Nirav
Chang, Eun
Liu, Yue
Metze, Florian
Stark, Tammy
Feizollahi, Zhaleh
Jessee, Andrea
Pujari, Mangesh
Aly, Ahmed
Damavandi, Babak
Wanga, Rakesh
Kumar, Anuj
Patel, Rohit
Yih, Wen-tau
Dong, Xin Luna
Computer Vision and Pattern Recognition
Wearable devices such as smart glasses are transforming the way people interact with their surroundings, enabling users to seek information regarding entities in their view. Multi-Modal Retrieval-Augmented Generation (MM-RAG) plays a key role in supporting such questions, yet there is still no comprehensive benchmark for this task, especially regarding wearables scenarios. To fill this gap, we present CRAG-MM -- a Comprehensive RAG benchmark for Multi-modal Multi-turn conversations. CRAG-MM contains a diverse set of 6.5K (image, question, answer) triplets and 2K visual-based multi-turn conversations across 13 domains, including 6.2K egocentric images designed to mimic captures from wearable devices. We carefully constructed the questions to reflect real-world scenarios and challenges, including five types of image-quality issues, six question types, varying entity popularity, differing information dynamism, and different conversation turns. We design three tasks: single-source augmentation, multi-source augmentation, and multi-turn conversations -- each paired with an associated retrieval corpus and APIs for both image-KG retrieval and webpage retrieval. Our evaluation shows that straightforward RAG approaches achieve only 32% and 43% truthfulness on CRAG-MM single- and multi-turn QA, respectively, whereas state-of-the-art industry solutions have similar quality (32%/45%), underscoring ample room for improvement. The benchmark has hosted KDD Cup 2025, attracting about 1K participants and 5K submissions, with winning solutions improving baseline performance by 28%, highlighting its early impact on advancing the field.
title CRAG-MM: Multi-modal Multi-turn Comprehensive RAG Benchmark
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2510.26160