Competition 1: The Challenge on Deepfake Detection and Localization

Short description:
As Deepfake technology continues to evolve, the proliferation of manipulated audio-visual content poses significant threats to societal trust and media integrity. The Deepfake Detection and Localization (DDL) Challenge invites global participants to address these pressing issues through three key goals: (1) enhancing detection interpretability by offering spatial-temporal localization of forgeries (e.g., pixel-level tampering, forged timestamps); (2) mitigating multi-modal threats by tackling complex cross-modality attacks such as “forged audio with authentic video”; and (3) promoting inclusivity and reproducibility by releasing the world’s largest multi-modal deepfake dataset, featuring over 1.8 million samples and 88 diverse forgery techniques. Join us in pushing the boundaries of trustworthy media forensics.

Website link:
https://deepfake-workshop-ijcai2025.github.io/main/index.html

Competition timeline:

Competition announcement & registration opens: March 15, 2025
Training data & validation data released: March 28, 2025
Test data released & registration closed: May 12, 2025
Test period ends: May 19, 2025
Submission of solutions and papers: May 26, 2025

Competition 2: MGTV Large Models-Driven Video Intelligence Challenge

Short description:
Large language models are revolutionizing every stage of the media industry pipeline. The MGTV Large Models-Driven Video Intelligence Challenge explores cutting-edge applications of LLMs in media understanding and user modeling. Track A focuses on deep semantic parsing, aiming to transform colloquial expressions and culture-specific speech content into precise textual representations. Track B emphasizes multimodal user behavior prediction, leveraging video, audio, and metadata to forecast user preferences more accurately. By constructing a complete technical closed-loop of “semantic understanding to scenario-based application,” this challenge addresses key bottlenecks in content creation and intelligent dissemination. We warmly invite developers, research teams, and enterprises worldwide to join us in accelerating the intelligent evolution of the content industry.

Website link:
https://challenge.ai.mgtv.com/#/home

Competition timeline:

Competition announcement date: May 15, 2025
User registration date: May 15, 2025
Competition start date: May 15, 2025
Team formation deadline: July 15, 2025
Submission deadline: July 18, 2025
Final results announcement date: July 30th,2025:

Competition 3: 1st AI for Drinking Water Chlorination Challenge

Short description:
Water distribution systems are essential infrastructures that ensure safe and reliable delivery of treated drinking water. To maintain water quality and safeguard public health, utilities commonly use chlorine as a disinfectant. However, rapid urbanization and climate change have introduced new complexities to water quality dynamics, challenging traditional chlorination control methods. This competition explores the potential of artificial intelligence—particularly reinforcement learning (RL)—to enhance chlorination strategies in response to wastewater contamination events. Participants are tasked with controlling chlorine injection in the CY-DBP water distribution system model, relying solely on data from strategically placed hydraulic and chlorine sensors. The goal is to develop AI-based control mechanisms that are both adaptive and resilient in real-world contamination scenarios. Join us in shaping the future of intelligent water infrastructure management.

Website link:
https://github.com/WaterFutures/AI-for-Drinking-Water-Chlorination-Challenge-IJCAI-25

Competition timeline:

Competition start: May 5, 2025
Submission deadline: August 1, 2025
Notification of results and publication of test scenarios: August 8, 2025
Presentation of results at IJCAI 2025: August 16–22, 2025 (Montreal)

Competition 4: The First SpikeCV Competition: Infinity – Wuji

Short description:
The First SpikeCV Competition explores the frontiers of neuromorphic visual computing with spike cameras—sensors capable of capturing ultra-high-speed, temporally continuous visual information. Centered on the theme Wuji, symbolizing boundlessness in speed and continuity, this competition challenges participants to redefine AI perception in extreme and dynamic environments. The main track focuses on high-speed texture reconstruction under complex scenarios, while two additional tracks promote the development of novel spike-based vision tasks and the design of open datasets and benchmarks. We invite researchers, engineers, and innovators worldwide to join this journey in advancing the next generation of bio-inspired vision systems.

Website link:
https://spikecv.github.io/competition

Competition timeline:

Competition announcement & registration opens: April 18, 2025
Submission begins (Track 1 ranking starts): May 1, 2025
Submission deadline for all tracks (code/data + demo video): July 10, 2025
Results announced: July 25, 2025
Award ceremony: August 28–31, 2025

Competition 5: Robust AI-driven Identification and Quantification of Medical Imaging Biomarkers: Pulmonary Fibrosis Segmentation Challenge 2025 (PuFiSeg2025)

Short description:
Pulmonary fibrosis encompasses a group of progressive lung diseases characterized by lung stiffening and scarring, with a mortality rate comparable to that of breast cancer—making it a critical global health concern. Accurate identification of fibrotic regions in high-resolution CT (HRCT) scans remains a major challenge due to factors such as blurred boundaries, inter/intra-observer variability, and the need for consistency across hundreds of 3D slices. Manual annotation is time-consuming and expertise-intensive. To address this, the PuFiSeg2025 Challenge introduces the first large-scale effort to automate pulmonary fibrosis segmentation. The dataset includes 262 expertly annotated 3D HRCT cases, with slice-by-slice labels curated by three experienced radiologists. Participants are encouraged to develop robust and generalizable models for automatic segmentation. This is an open-call challenge, and post-conference submissions will also be accepted to continue advancing the field.

Website link:
https://www.codabench.org/competitions/7330/

Competition timeline:

Challenge website & registration opening: April 20, 2025
Training cases released: June 5, 2025
Validation cases released: June 20, 2025
Test phase opening (Docker submissions): July 30, 2025
All submissions closed: August 16, 2025

Competition 6: 1st Challenge on 4D Micro-Expression Recognition for Mind Reading (4DMR 2025)

Short description:
The 4DMR 2025 Challenge pioneers the task of 4D micro-expression recognition by leveraging high-resolution 3D facial mesh sequences from the newly released 4DME dataset. Participants are invited to develop models that can accurately detect subtle, transient facial expressions across five emotional categories: Negative, Positive, Repression, Surprise, and Others. Unlike traditional 2D video-based methods, this challenge utilizes temporal 3D (i.e., 4D) data to capture the intricate dynamics of facial muscle movement, offering a more realistic and challenging benchmark for spontaneous emotion recognition. Framed as a multi-label classification task, the challenge reflects the inherently complex and overlapping nature of human emotional expression. Emphasizing cross-cultural generalization, fairness, and robustness, this competition aims to advance the frontiers of affective computing and mind-reading AI.

Website link:
https://www.kaggle.com/competitions/4-dmr-workshop-challenge-2025-privateenience2

Competition timeline:

Registration opens: April 20, 2025
Challenge start / dataset release: April 22, 2025
Submission deadline: June 28, 2025
Evaluation period: June 29 – July 5, 2025
Final results announced: July 10, 2025
Workshop & presentations: During IJCAI 2025 (exact date TBD)

Competition 7: Barkopedia: A Canine Vocalization Understanding Challenge

Short description:
Dogs vocalize to express emotions, intentions, and situational cues—but understanding these vocal signals remains a complex task. Barkopedia, hosted at IJCAI 2025, introduces a large-scale, real-world challenge in canine vocalization analysis. The competition addresses key limitations in previous studies—such as small datasets and handcrafted features—by offering a rich, diverse corpus of dog barks collected from YouTube and Reddit. The challenge comprises eight tasks, each designed to push forward the frontiers of machine learning, audio processing, and animal communication. All competition components, including dataset access, leaderboards, and result submissions, are hosted on Hugging Face. Up to 20 teams per task are invited, with the option to enter multiple tasks and submit multiple runs. Top teams will be recognized with certificates and invited to present their solutions at IJCAI 2025.

Website link:
https://uta-acl2.github.io/Barkopedia/

Competition timeline:

Competition opens for registration and training data release: April 15, 2025
Validation data and baseline results released: June 1, 2025
Registration deadline: June 15, 2025
Submission portal opens for test data: July 1, 2025
Final submission deadline: July 15, 2025
Preliminary results announced; feedback period opens: August 1, 2025
Deadline for challenge reports and code submission: August 15, 2025
Final results and winner announcement: August 30, 2025
Presentation of winning solutions at IJCAI 2025: September 15, 2025

Competition 8: Counterfactual Routing Competition

Short description:
Traditional path planners like Dijkstra’s algorithm compute optimal routes but offer little insight into the reasoning behind specific routing decisions. The Counterfactual Routing Competition introduces a novel task focused on explainable personalized routing. Participants are provided with a map, a start and destination, a route planner, and a user model encoding personalized preferences. From these inputs, the optimal route—called the fact route—is generated. Additionally, a suboptimal alternative—the foil route—is also given. The user asks: “Why is the fact route better for me than the foil route?” The challenge is to generate a counterfactual map, minimally altered from the original, under which the foil route would be optimal. This task promotes transparency in AI decision-making for real-world navigation systems.

Website link:
https://sites.google.com/view/crc25-ijcai/home

Competition timeline:

Full competition registration open: May 1, 2025
Competition submission deadline: July 10, 2025
Competition results release: July 30, 2025
Announcement of presentation slots: August 5, 2025
Results presentation and winner announcement: During IJCAI 2025

Competition 9: The Fifth Mahjong AI Competition

Short description:
The Fifth Mahjong AI Competition invites participants to develop intelligent Mahjong agents capable of competing against other AIs and human players on Botzone, a widely used online AI competition platform. The competition adopts the Mahjong Competition Rules (MCR), also known as Chinese Official Mahjong. To assist with development and debugging, a sample game-playing agent and a judge program are provided. Participants will be ranked based on their performance across two official rounds: a tournament round and a final round. Success in this competition may benefit from integrating various AI techniques, including planning, game theory, heuristic search, Monte Carlo Tree Search (MCTS), opponent modeling, and machine learning. Join us in pushing the boundaries of strategic multi-agent gaming.

Website link:
https://botzone.org.cn/static/gamecontest2025a.html

Competition timeline:

Registration deadline: June 9, 2025 (GMT+8)
Round 0 (Simulation Round): ~June 9, 2025 (GMT+8)
Round 1 (Tournament Round): June 9, 2025 (GMT+8)
Round 1 results announced: June 16, 2025 (GMT+8)
Round 2 (Final Round): July 7, 2025 (GMT+8)
Final results announced: During IJCAI 2025

Competition 10: Social Media Prediction Challenge

Short description:
The Social Media Prediction (SMP) Challenge is an annual competition inviting researchers, engineers, and practitioners to tackle cutting-edge AI tasks related to social multimodal content. In the era of explosive social media growth—ranging from image-centric posts to short-form videos and memes—understanding and predicting user engagement has become crucial. This challenge introduces a large-scale, richly annotated Social Media Prediction Dataset (SMPD), designed to support robust and generalizable models for forecasting user behavior. Participants are encouraged to develop novel techniques that enhance predictive accuracy, sociological understanding, and real-world applicability. The competition ultimately aims to drive progress in modeling dynamic social interactions, benefiting creators, platforms, and end-users alike.

Website link:
https://www.smpv-challenge.com

Competition timeline:

SMP Challenge open: April 20, 2025
Submission open: April 20, 2025
Submission deadline: June 15, 2025
Evaluation and reproducibility review: June 15–25, 2025
Award notification: July 20, 2025

Competition 11: The 2nd Challenge on Characterizing User Behavior in Social Networks: Propagation, Prediction, and Sensemaking

Short description:
Social media platforms function as complex, graph-structured ecosystems that facilitate modern communication and social interaction. These dynamic networks—composed of evolving nodes (users, content) and edges (interactions)—pose unique challenges for modeling user behavior and information diffusion. This competition bridges graph machine learning with social computing, inviting interdisciplinary approaches that span network science, behavioral psychology, and cybersecurity. The challenge focuses on two interconnected tasks: (1) retweet network link prediction and (2) bursty event cascade classification. Together, these tasks aim to enhance graph-structured reasoning capabilities, promote better understanding of subgraph patterns, and foster human-centric insights into user behavior on social platforms. By modeling relational dynamics and temporal bursts in social cascades, participants will contribute to advancing interpretable and predictive models for digital social ecosystems.

Website link:
https://magic-group-buaa.github.io/IJCAI25/index.html

Competition timeline:

Data release, platform setup, and participant registration: April 16, 2025
Final solution submission deadline: July 20, 2025
Winning teams announced: August 10, 2025

Competition 12: The 4th RePSS – Multimodal Fusion Learning for Remote Physiological Signal Sensing

Short description:
Facial videos contain subtle color changes caused by cardiac pulsations which can be reconstructed as remote photoplethysmography (rPPG). rPPG signals contain physiological information such as the heart rate (HR), heart rate variability (HRV), breathing rate (BR), and oxygen saturation, which are important indicators of people’s health and psychological status. The Remote Physiological Signal Sensing (RePSS) Challenge aims to gather researchers from different disciplines to promote rPPG measurement research. RePSS is organized as an event series, and this year the 4th RePSS will be held in conjunction with IJCAI 2025, focusing on multimodal fusion learning for rPPG measurement from multiple sensors.

Website link:
https://kaggle.com/competitions/the-4th-repss-t1

Competition timeline:

Challenge website online: March 24,2025
Release training and resource data: April 1, 2025
Team registration deadline: April 30, 2025
Testing data release: May 3, 2025
Final test submission deadline: May 17, 2025
Challenge results release: May 24, 2025
Paper submission deadline: May 31, 2025
Notification to authors: June 6, 2025

Competition 13: The 16th Automated Negotiating Agents Competition (ANAC 2025)

Short description:
The Automated Negotiating Agent Competition (ANAC) brings together researchers from the negotiation community and provides a unique benchmark for evaluating practical negotiation strategies in multi-issue domains. ANAC has the following aims to provide an incentive for the development of effective and efficient negotiation protocols and strategies for bidding, accepting, and opponent modeling for different negotiation scenarios; to collect and develop a benchmark of negotiation scenarios, protocols, and strategies; to develop a standard set of tools and criteria for the evaluation and exploration of new protocols and new strategies against benchmark scenarios, protocols, and strategies and to set the research agenda for automated negotiation. This year, we have

Automated Negotiation League: Designing and building a negotiation agent for multi-deal one-to-many negotiation. The agent encounters multiple opponents in sequence.
Supply Chain Management League: Designing and building an autonomous agent that negotiates for a factory manager in a supply chain management simulation.

Website link:
https://web.tuat.ac.jp/~katfuji/ANAC2025/

Competition timeline:

Registration on the competition website (Recommended): April 25th, 2025
Academic Report submission deadline: June 5th, 2025
Submission deadline for the competition: 1 June, 2025
Notification of the finalists: 15 June, 2025
Competition Session: 16-22 August, 2025

Competition 14: Autonomous Travel Itinerary Planning Challenge

Short description:
The IJCAI-25 Travel Planning Challenge aims to bridge this gap by fostering innovations in AI agents capable of generating robust, personalized travel plans. Participants will tackle the task of synthesizing detailed itineraries—spanning transportation, accommodation, activities, and cost management—based on user queries with heterogeneous requirements. This competition intersects cutting-edge domains such as Natural Language Processing, LLM Reasoning and Planning, Constraint Optimization, and Neuro-Symbolic Learning, offering a unique platform to explore the synergy of generative AI, recommendation systems, and planning technologies.

Website link:
https://chinatravel-competition.github.io/IJCAI2025/

Competition timeline:

Baseline Code & Data available; Registration Begins: May 10, 2025
Phase 1 Results submission deadline: July 31, 2025
Phase 2 results submission: August 1, 2025
Technique report submission: August 30, 2025