Accepted Grand Challenges
Group 1: Deepfake Detection & AI-Generated Media
Challenge 1
Title: (MillionDeepfake) The 2nd One Million Deepfakes Detection Challenge
URL: https://deepfakes1m.github.io/2025
Abstract: This ACM Multimedia 2025 challenge builds upon the successful 2024 Grand Challenge hosted at ACM Multimedia 2024, which attracted participation from 194 teams. The challenge focusses on the tasks of detection and localisation of deepfakes. The later deals with identifying the timestamps where a video is manipulated. This year the challenge is based on AV-Deepfakes-1M++ database.
Evaluation results deadline: 27 June 2025
Paper submission deadline(on invitation): 01 August 2025
Challenge 2
Title: (AADD-2025) Adversarial Attacks on Deepfake Detectors: A Challenge in the Era of AI-Generated Media
URL: https://iplab.dmi.unict.it/mfs/acm-aadd-challenge-2025/
Abstract: The aim of this challenge is to expose and address vulnerabilities in current deepfake detection systems by designing adversarial attacks that alter deepfake images—rendering them unrecognizable as synthetic content to 4 proposed classifiers—while preserving high visual similarity to the original images. Participants will be provided with one dataset, divided into sixteen subsets: four GAN-based models and four diffusion-based models in high quality resolution, and four GAN-based models and four diffusion-based models in low quality resolution. Participants must focus on the entire dataset. In addition to developing effective adversarial perturbations, participants are required to submit an abstract that outlines their methodology and attach the adversarial images.
Registration deadline: 22 May 2025
Evaluation results deadline: 15 June 2025
Paper submission deadline(on invitation): 30 June 2025
Challenge 3
Title: (IPVG) Identity-Preserving Video Generation Challenge
URL: https://hidream-ai.github.io/ipvg-challenge.github.io/
Abstract: In this grand challenge, we introduce Identity-Preserving Video Generation (IPVG) task, which maintains the consistency of given reference identity along text-to-video generation process. This year’s IPVG grand challenge includes two tracks: Facial Identity-Preserving Video Generation and Full-body Identity-Preserving Video Generation. In general, the goal of this grand challenge is two-fold: (a) coalescing community effort around a new challenging identity-preserving video generation dataset, and (b) offering a fertile ground for designing controllable generative models to facilitate precise identity binding in video generation, aiming to propel the field toward more accountable and user-steerable video synthesis systems.
Evaluation results deadline: 20 June 2025
Paper submission deadline(on invitation): 05 July 2025
Challenge 4
Title: (I2V Model Acceleration) Image-to-Video Generation Model Acceleration Challenge
URL: https://challenge.ai.mgtv.com/#/track/53
Abstract: Video generation has emerged as a pivotal technology in multimedia applications, yet its widespread adoption is hindered by the computational inefficiency of current models. This challenge focuses on accelerating domain-specific video generation systems through innovations in algorithmic optimization. By enhancing inference efficiency while preserving output quality, the Challenge seeks to address critical barriers in deployment scalability and operational costs. Successful solutions could empower resource-constrained developers to leverage advanced video synthesis capabilities through optimized computational paradigms, fostering broader industrial adoption and advancing sustainable AI development in multimedia ecosystems.
Evaluation results deadline: 25 July 2025
Paper submission deadline(on invitation): 30 July 2025
Group 2: Multimodal Large Vision-Language AI & Content Generation
Challenge 5
Title: (LAVA Challenge) Large Vision – Language Model Learning and Applications Grand Challenge
URL: https://lava-workshop.github.io/grandchallenge
Abstract: Recent advances in large vision – language models (LVLMs) hold immense promise across various domains, including healthcare, education, entertainment, transportation, and finance, by enabling more sophisticated and context-aware multimedia interactions. Despite their potential to advance multimodal understanding, LVLMs pose significant research challenges, especially in understanding business-related documents. In order to enhance the ability of Large Vision-Language Models (LVLMs) to accurately interpret and generate descriptive text from complex visual inputs in business-related documents, we propose this grand challenge which is planned to hold in consecutive years.
In this challenge, we focus on Q&A tasks including multiple-choice and opened questions on data including Graphs, Charts, Diagrams, Data Flow Diagrams (DFDs), Class Diagrams, Gantt Charts, and Building Design Drawings. Participants can register as individuals or teams. They are required to develop a model that can answer questions related to the input data.
Registration deadline: 31 May 2025
Evaluation results deadline: 30 June 2025
Paper submission deadline(on invitation): 30 June 2025
Challenge 6
Title: (MIRAGE) Multimodal Interleaved Reasoning and Generation Challenge
URL: https://mm25mirage.github.io/mirage/
Abstract: MIRAGE aims to advance multimodal AI systems’ ability to handle interleaved vision-language tasks through two sub-tracks:
Track A: Multimodal Interleaved Instruction Reasoning
Focuses on understanding relationships between multiple images and text instructions, enabling systems to answer open-ended or multi-choice questions requiring cross-modal reasoning.
Track B: Multimodal Interleaved Content Generation
Emphasizes generating contextually consistent images based on sequential interleaved references (text and images), ensuring alignment with instruction semantics.
Evaluation results deadline: 20 June 2025
Paper submission deadline(on invitation): 10 July 2025
Challenge 7
Title: (IntentVC) Intention-Oriented Controllable Video Captioning
URL: https://sites.google.com/view/intentvc/
Abstract: The IntentVC Challenge tackles a fundamental limitation in video captioning by introducing user-controllable, intention-oriented outputs. Unlike traditional approaches that generate general-purpose, one-size-fits-all descriptions of videos, this challenge enables captions tailored to specific user-defined intentions that can test video understanding modules in capturing various aspects of the video.
Evaluation results deadline: 20 June 2025
Paper submission deadline(on invitation): 17 July 2025
Challenge 8
Title: (ResMM) Truthful and Responsible Multimodal Learning Challenge
URL: https://mm-hall-fact.github.io/ACMMM2025/
Evaluation results deadline: 20 June 2025
Paper submission deadline(on invitation): 15 June 2025
Group 3: Multimodal Sentiment, Emotion & Personality Analysis
Challenge 9
Title: (MER2025) Semi-supervised, Fine-grained, Descriptive Emotion Understanding with Emotion-enhanced Personality Recognition
URL: https://zeroqiaoba.github.io/MER2025-website/
Abstract: Multimodal emotion recognition is an active research topic in artificial intelligence. Its main goal is to integrate multi-modalities to identify human emotional states. Current works generally assume accurate emotion labels for benchmark datasets and focus on developing more effective architectures. However, existing technologies are difficult to meet the demand for practical applications. To this end, we continuously organize MER23 Challenge@ACM MM, MRAC23 Workshop@ACM MM, MER24 Challenge@IJCAI, and MRAC24 Workshop@ACM MM. In this year, we will continuously holding related workshops and challenges that bring together researchers around the world to further discuss recent research and future directions for robust multimodal emotion recognition.
Evaluation results deadline: 10 July 2025
Paper submission deadline(on invitation): 20 July 2025
Challenge 10
Title: (AVI2025) Assessing Personality Traits and Interview Performance from Asynchronous Video Interviews
URL: https://avichallenge.github.io/
Abstract: Asynchronous Video Interviews (AVIs) allow candidates to record responses to predefined questions using digital devices, offering flexibility and enabling remote assessments. Evaluating personality traits and interview performance through AVIs provides organizations with valuable insights into candidates’ profiles and helps predict future job performance. We are hosting the ACM MM 2025 AVIs challenge to promote advancements in personality traits and interview performance assessment at ACM MM 2025. This event brings together researchers from psychology, artificial intelligence and affective computing to explore new multimedia AI models aligning with real-life AVI practices for job-related assessment.
We introduce a novel dataset of mock AVIs (3876 videos from 646 subjects) conducted in a simulated job application procedure. Interview questions were designed by psychologists to reflect real-world selection contexts and activate personality traits based on Trait Activation Theory. Annotations for personality traits and job competencies were provided by trained evaluators and professional recruiters, ensuring methodological and ecological validity.
Evaluation results deadline: 01 July 2025
Paper submission deadline(on invitation): 12 July 2025
Challenge 11
Title: (MPDD 2025) 1st MPDD Challenge: Multimodal Personality-aware Depression Detection
URL: https://hacilab.github.io/MPDDChallenge.github.io/#HOME
Abstract: Depression is a pervasive mental health disorder with a profound impact on global well being , affecting millions and significantly hindering quality of life, productivity, and overall health. Many datasets primarily focus on establishing direct correlations between behaviora l signals and depression levels. This approach, while valuable, often fails to account for the nuanced individual variations in depressive symptoms. Moreover, the demographic scope of existing datasets is predominantly confined to young adults, thereby overlooking other critical age groups, such as the elderly, who may present distinct patterns of depression.
The Multimodal Personality-aware Depression Detection (MPDD)aims to address these gaps by integrating multiple data types (audio, video, text) with individual difference information such as personality traits, age, gender, and health conditions. This challenge seeks to develop models that can accurately detect depression across a broader demographic, fostering more inclusive and effective mental health detection systems.
The advent of specialized datasets has been instrumental in advancing depression re search by providing multimodal data that can enhance the detection and understanding of depressive symptoms. For instance, the AVEC challenge has contributed rich audiovisual interview data annotated with depression-related information, facilitating more nuanced assessments. The DAIC-WoZ dataset has furthered our understanding by incorporating clinical interviews and PHQ-9 scores. Other datasets, including the Pitts burgh dataset, the D-Vlog dataset, the EATD-Corpus, the CMDC, and the MODMA, have expanded the scope by integrating diverse data types like EEG and utilizing standard depression scales for evaluation. Additionally, datasets such as AMI GOS and DEAP have incorporated physiological data, offering insights into the physiological underpinnings of depression, while the SEED dataset provides EEG and eye movement data annotated with personality traits, highlighting the role of individual differences in depressive symptoms.
To advance the development of personalized and accurate depression detection systems, we are launching the MPDD 2025: Multimodal Personality-aware Depression Detection.This challenge aims to push the boundaries of depression detection by integrating multimodal data (audio, video, text) with individual difference information such as personality traits, age, gender, and health conditions.
Evaluation results deadline: 15 June 2025
Paper submission deadline(on invitation): 30 June 2025
Challenge 12
Title: (MEGC) Facial Micro-Expression Grand Challenge
URL: https://megc2025.github.io/
Abstract: Facial micro-expressions (MEs) are involuntary facial movements that occur when an individual experiences an emotion but attempts to suppress it, often in high-stakes situations. ME analysis is a growing interdisciplinary field that has recently expanded into multimodal approaches. The Facial Micro-Expression Grand Challenge (MEGC) 2025 introduces two frontier tasks: a Spot-Then-Recognize task, which aims to accurately classify expression sequences following the spotting step in a sequential manner, and a Visual Question Answering task, which leverages advanced vision-language and multimodal large language models to infer ME attributes, recognize emotions, and enable natural language interactions. This challenge seeks to drive innovation in ME analysis by fostering new research directions and establishing benchmark evaluations, ultimately advancing human-centered multimedia applications.
Evaluation results deadline: 27 June 2025
Paper submission deadline(on invitation): 30 July 2025
Challenge 13
Title: (REACT 2025 Challenge) The Third Multiple Appropriate Facial Reaction Generation in Dyadic Interactions
URL: https://sites.google.com/view/react2025
Abstract: The REACT 2025 third Multiple Appropriate Facial Reaction Generation (MAFRG) in Dyadic Interactions to be held in conjunction with the ACM Multimedia (ACM-MM) 2025 in Dublin, Ireland, is calling for contributions in the form of papers and competition in the challenge. The competition will be using the first dataset that is specifically collected for MAFRG tasks, theMulti-modal Multiple Appropriate Reaction in Social Dyads (MARS) Dataset, which records audio, visual and EEG signals for each participant.
Evaluation results deadline: 26 June 2025
Paper submission deadline(on invitation): 30 June 2025
Challenge 14
Title: (MERG) Avatar-based Multimodal Empathetic Conversation Challenge
URL: https://avamerg.github.io/MM25-challenge
Abstract: The Avatar-based Multimodal Empathetic Conversation (MERG) challenge falls within the research domain of multimodal affective computing and empathetic dialogue generation. Unlike traditional text-only empathetic response tasks, the MERG challenge requires models to generate contextually appropriate, emotionally aligned responses in a multimodal setting. Given a multimodal dialogue context, models are expected to generate an empathetic response that includes three synchronized components: textual reply, emotive speech audio, and expressive talking face video. To support this challenge, a large-scale MERG dataset will be provided, consisting of high-quality multimodal dialogue-response pairs. The dataset includes diverse conversational scenarios where human annotators have carefully crafted emotionally rich responses, ensuring alignment across text, speech, and visual expressions. The challenge is struc- tured into three subtasks: (1) Text-Only Empathetic Response Gen- eration, which focuses on generating an empathetic textual reply based solely on the dialogue history; (2) Multimodal-Aware Em- pathetic Response Generation, which requires generating a tex- tual response while leveraging multimodal contextual information from speech and visual cues; and (3) Multimodal Empathetic Re- sponse Generation, which involves generating a fully multimodal response, including synchronized text, speech, and facial expres- sions, to achieve a natural and emotionally aligned interaction.
Evaluation results deadline: 20 June 2025
Paper submission deadline(on invitation): 20 July 2025
Group 4: Multimedia Verification & Social Media Misinformation Detection
Challenge 15
Title: (MV) Grand Challenge on Multimedia Verification
URL: https://multimedia-verification.github.io
Abstract: The Grand Challenge on Multimedia Verification is an open competition where researchers and practitioners compete to verify the authenticity and context of multimedia content, addressing real-world misinformation challenges. Participants analyze provided images and videos, assess sources, detect manipulations, and submit a detailed verification report.
Evaluation results and paper deadline: 13 June 2025
Challenge 16
Title: (SMP Challenge) Social Media Prediction Challenge
URL: https://smp-challenge.com/
Abstract: SMP Challenge is an annual challenge that seeks excellent research teams on new ways of forecasting problems and meaningfully improving people’s social lives and business scenarios. The enormous amounts of online content lead to overconsumption, online word-of-mouth helps us to efficiently discover interesting news, emerging topics, the latest stories, or amazing products from the information ocean. Therefore, predicting online popularity became an emerging and significant task for online media, brand marketing, social influencers, or our individuals. The challenge task Social Media Popularity Prediction (SMPP) aims at predicting the impact of online post sharing on social media. It is central to various scenarios, such as online advertising, social recommendation, demand forecasting, etc.
Evaluation results deadline: 25 May 2025
Paper submission deadline(on invitation): 01 July 2025
Challenge 17
Title: (NewsEvent) News Event Retrieval and Explanation from Large Collection of TV News Videos
URL: https://aichallenge.hcmus.edu.vn/acm-mm-2025/newsretrieval
Abstract: The News Event Retrieval and Explanation Grand Challenge at ACM Multimedia 2025 is focused on retrieving, analyzing, and explaining news events from large-scale TV news video collections.
The challenge features two evaluation modes: Interactive Mode (similar to VBS and LSC), where users refine queries and interact with the system, and Automatic Mode, where AI autonomously retrieves and explains news events. In Interactive Mode, the focus is on real-time human-AI collaboration, while Automatic Mode tests autonomous retrieval and reasoning. Both modes assess efficiency, accuracy, and scalability.
Participants will work with a large, multimodal dataset of TV news videos from various sources, including CNN, BBC, Reuters, and Vietnamese news channels. The dataset will include annotations such as speech transcriptions, OCR-extracted text, and temporal event timestamps. The challenge will use AI models for multimodal retrieval, video segmentation, and event explanation.
Registration deadline: 15 June 2025
Evaluation results deadline: 01 July 2025 – 30 July 2025
Paper submission deadline(on invitation): 01 August 2025
Group 5: Human-Robot & AI Conversational Understanding
Challenge 18
Title: (McABSA) Multimodal Conversational Aspect-based Sentiment Analysis Challenge
URL: https://panosent.github.io/MM25-challenge
Abstract: We propose the Multimodal Conversational Aspect-based Sentiment Analysis (MCABSA) challenge to the community of aspect- based sentiment analysis. This challenge falls within the research domain of sentiment analysis and multimodal learning. In the MCABSA challenge, models and systems are expected to address two novel subtasks: 1) Panoptic Sentiment Sextuple Extraction, panoramically recognizing holder, target, aspect, opinion, sentiment, and rationale from multi-turn, multi-party multimodal dialogue. 2) Sentiment Flipping Analysis, detecting the dynamic sentiment transformation throughout the conversation along with the causal reasons. Alongside the challenge, we introduce the PanoSent dataset, a high-quality large-scale resource annotated both manually and automatically. PanoSent features multimodality, multi-scenario coverage, and includes both implicit and explicit sentiment elements, thereby offering a comprehensive testbed for diverse real-world scenarios.
Registration deadline: 20 May 2025
Evaluation results deadline: 15 June 2025
Paper submission deadline(on invitation): 15 July 2025
Challenge 19
Title: (ERR@HRI 2.0) Multimodal Detection of Errors and Failures in Human-Robot Conversations
URL: https://sites.google.com/view/errhri20
Abstract: While the integration of large language models (LLMs) into conversational robots has made human-robot conversations more flexible, LLM-powered conversational robots remain prone to errors, such as misinterpreting user intent, interrupting users, or failing to respond. Detecting and addressing these failures is critical to preventing conversational breakdowns, task disruptions, and maintaining user trust. Towards tackling this challenge, the ERR@HRI 2.0 challenge provides a multimodal dataset of LLM-powered conversational robot failures during human-robot conversations and encourages researchers to benchmark machine learning-based models designed to identify when robot failures occur. This dataset features multimodal interaction data, including facial, speech, and head features from 15 hours of dyadic human-robot conversations multimedia, annotated with labels indicating the presence or absence of robot errors and user intent to correct for a mismatch between robot behavior and their expectation. We invite participants to collaborate in teams to develop and submit multimodal ML models designed to detect conversational robot failures, which will be benchmarked based on various performance metrics, including accuracy, precision, recall, F1 score, and timing-based metrics.
Evaluation results deadline: 30 June 2025
Paper submission deadline(on invitation): 28 July 2025
Challenge 20
Title: (MultiMediate’25) Cross-cultural Multi-domain Engagement Estimation
URL: https://multimediate-challenge.org/
Abstract: Estimating the momentary level of engagement from multi-modal participant behaviour is an important prerequisite for assistive systems that support human interactions. At the same time, the connection between engagement and behaviour heavily depends on situational and cultural characteristics. Therefore, MultiMediate’25 poses the cross-cultural, multi-domain engagement estimation challenge. The challenge covers a unique combination of training and test data from diverse cultural backgrounds, including Japanese, Chinese, German, Arabic, Indonesian, and French speakers, as well as different interaction situations (group vs. dyadic). In addition to evaluation metrics that quantify the overall quality of predictions, MultiMediate’25 will also quantify the amount of bias in predictions with respect to the variables gender and cultural background. In addition to engagement estimation, we continue to invite submissions to popular tasks from previous iterations of the challenge: eye contact detection, bodily behaviour recognition, and backchannel detection.
Evaluation results deadline: t.b.a.
Paper submission deadline(on invitation): 30 June 2025
Group 6: Biomedical & Scientific Image Processing
Challenge 21
Title: (HeNeCT) Segmentation and Description of Space-occupying Lesions in Head & Neck CT Scan
URL: https://aichallenge.hcmus.edu.vn/acm-mm-2025/henect
Abstract: The HeNeCT Grand Challenge at ACM Multimedia 2025 aims to advance the segmentation and description generation of space-occupying lesions in head and neck CT scans. The challenge introduces two main tasks:
- Lesion Segmentation – Participants develop deep learning models to accurately segment tumors and abscesses in contrast-enhanced head and neck CT scans.
- Lesion Description Generation – Participants develop models to generate structured, radiology-style reports for detected lesions, capturing lesion type, size, shape, and other critical features.
- The dataset for the challenge will be annotated by expert radiologists and will include files with tumor and abscess cases. The data will be released in three stages, progressively increasing in complexity. Evaluation metrics for segmentation include IoU, DSC, and HD95, while evaluation for description generation includes BLEU, ROUGE, METEOR, and CIDEr.
Evaluation results deadline: t.b.a.
- Paper submission deadline(on invitation): t.b.a.
Challenge 22
Title: (ENTRep) Advancing Vision-Language AI for ENT Endoscopy Analysis
URL: https://aichallenge.hcmus.edu.vn/acm-mm-2025/entrep
Abstract: The ENTRep Challenge focuses on advancing AI techniques for ENT endoscopy analysis. Tasks include automated lesion detection and classification, medical report generation, visual question answering (VQA), and text-to-image generation for ENT pathologies. The goal is to enhance diagnostic accuracy and optimize clinical workflows using AI. The challenge will provide a high-quality, annotated dataset of ENT endoscopy images and a multi-stage training approach, starting with basic tasks and progressing to more complex scenarios. Evaluation will focus on AI accuracy in classification, detection, segmentation, and report generation, along with expert clinical evaluation. The ENTRep Challenge tasks include:
- Image Classification: Classifying ENT endoscopy images by anatomical region and pathology.
- Lesion Detection & Segmentation: Detecting and segmenting lesions in ENT images.
- Medical Report Generation: Automatically generating structured clinical reports from images.
- Visual Question Answering (VQA): Answering medical questions based on ENT images.
- Text-to-Image Generation: Generating synthetic ENT images from text descriptions.
Registration deadline: 31 May 2025
Evaluation results deadline: 30 June 2025
Challenge 23
Title: (GroMo25) Plant Growth Modeling with Multiview Images
URL: https://data.annam.ai/gromo25/
Abstract: Understanding plant growth dynamics is essential for applications in agriculture and plant phenotyping. We present the Growth Modelling (GroMo) challenge, which is designed for two primary tasks: (1) plant age prediction and (2) leaf count estimation, both essential for crop monitoring and precision agriculture. For this challenge, we introduce GroMo25, a dataset with images of four crops: radish, okra, wheat, and mustard. Each crop consists of multiple plants (p1, p2, …, pn) captured over different days (d1, d2, …, dm) and categorized into five levels (L1, L2, L3, L4, L5). Each plant is captured from 24 different angles with a 15-degree gap between images. Participants are required to perform both tasks for all four crops with these multiview images. The GroMo Challenge aims to advance plant phenotyping research by encouraging innovative solutions for tracking and predicting plant growth.
Registration deadline: 30 April 2025
Evaluation results deadline: 13 June 2025
Paper submission deadline(on invitation): 30 June 2025
Group 7: Micro-Action, Small-Scale Behavior & Event Detection
Challenge 24
Title: (MAC 2025) The 2nd Micro-Action Analysis Grand Challenge
URL: https://sites.google.com/view/micro-action
Registration deadline: 25 June 2025
Evaluation results deadline: 08 July 2025
Paper submission deadline(on invitation): 19 July 2025
Challenge 25
Title: (EVENTA 2025) Event-Enriched Image Analysis Grand Challenge
URL: https://selab.hcmus.edu.vn/events/eventa
Abstract: Conventional image analysis tasks (e.g., image captioning) often provide superficial descriptions, focusing on objects and actions while missing deeper contextual details. These methods struggle to convey who is involved, what is happening, and the significance of events, limiting their ability to generate informative captions for real-world scenarios. The Event-Enriched Image Analysis (EVENTA) Grand Challenge enhances image-caption retrieval and generation by integrating event-related context. Capturing not just visual elements but also information from events in the real-world enables richer, narrative-driven captions that offer a more complete understanding of an image. These captions go beyond simple visual descriptions by offering deeper insights that cannot be gleaned from merely observing the image. This challenge promotes two tracks: Event-Enriched Image Captioning and Event-Based Image Retrieval.
Registration deadline: 09 June 2025
Evaluation results deadline: 24 June 2025
Paper submission deadline(on invitation): 01 July 2025
Challenge 26
Title: (CASTLE) 1st CASTLE Multimodal Analytics Challenge
URL: https://castle-dataset.github.io/
Abstract: The increasing availability of mobile computing devices equipped with a variety of sensors enables us to capture and quantify more and more aspects of the human condition. Automatically drawing insight from such captures has the potential to open doors to a range of new applications, but remains challenging, in particular, when information needs to be combined from different points along the timeline.
The CASTLE challenge aims to act as a catalyst in the development of methods for multimodal understanding by providing a rich multimodal dataset that serves as a basis for a range of analysis tasks. Research on lifelog retrieval and analysis has so far focused on longitudinal data of a single person, or multi-person data over a short time range. The CASTLE challenge scales the problem to multi-person and multi-day data, aiming to model real-world problem setting more closely, and advancing the state of the art in multimodal understanding of human activity videodata.
Evaluation results and paper deadline: 30 June 2025