Organizations: Ant Group Shanghai, China · Zhejiang University Hangzhou, China · Ant Group Beijing, China · Ant Group Hangzhou, China · The Chinese University of Hong Kong Hong Kong, China
The EgoLink Challenge at ACM Multimedia 2026 benchmarks egocentric multimodal intelligence linking first-person perception, social reasoning, and interactive action. Unlike benchmarks centered on visible objects and activities, EgoLink evaluates whether models can interpret emotional and social cues, resolve visually grounded references, and convert user requests into verified tool-mediated actions. Its two tracks cover social reasoning in egocentric video and interactive agents that complete service tasks through dialogue, tool use, and closed-loop execution. Together, they test whether perception, reasoning, and decision-making remain coupled under both offline question answering and state-changing interaction. We highlight a multi-stage Track 1 data pipeline for multiple-choice question (MCQ) generation, chain-of-thought generation, automatic quality grading, and video-contribution filtering. We summarize the task design, evaluation metrics, rules, leaderboards, and representative solutions. The benchmark combines objective answer scoring with process- and result-based evaluation, enabling analysis of both final correctness and execution quality. The best Track 1 system achieved 0.9263 accuracy, while the best Track 2 system achieved 0.844 joint success and 0.916 micro accuracy. Participant reports reveal a shared principle: strong systems make intermediate evidence explicit through candidate verification, option-robust reasoning, scenario-aware planning, tool canonicalization, and state-aware correction before answering or acting. These findings highlight the importance of process verification and evidence-grounded reasoning for reliable egocentric multimodal intelligence.
Figures & tables
Figure 1. EgoLink challenge overview.
Figure 2. Video-Contribution Filtering for Track 1. Candidates are screened to reduce questions answerable without visual evidence.
Metric
Description
Track 1 accuracy
Overall top-1 MCQ accuracy over the evaluation questions.
Dimension scores
Scores for emotion, causality, intent, and summarization question groups.
Track 2 joint success
Task-level success requiring both correct tool process and correct database result.
Micro accuracy
Call-level tool accuracy for diagnosing fine-grained execution behavior.
Result/tool success
Separate result-based and tool-based scores that expose process-result mismatches.
Table 1. Evaluation signals used in EgoLink.
Rank
Team
Accuracy
1
GXU-LIPE
0.9263
2
Lai
0.9246
3
TeleAI
0.9132
4
EgoMinds
0.9092
5
wjdd
0.9089
5
MR-CAS
0.9089
Table 2. Official Track 1 leaderboard excerpt. Accuracy is the overall top-1 MCQ score reported on the challenge website.
Team
Joint
Micro
Result
Tool
MediaLab
0.844
0.916
0.875
0.854
GML-MM Group
0.823
0.914
0.854
0.823
BigVisionAgentGoGoGo
0.781
0.889
0.896
0.802
uestc_goat
0.729
0.814
0.760
0.740
EgoMinds
0.604
0.733
0.646
0.615
Table 3. Official Track 2 leaderboard excerpt. Joint success is the primary ranking metric.
Team
Track
Main mechanism
Reported evidence
GXU-LIPE
Track 1
Question-conditioned social evidence graph, candidate-wise verification, source-traceable reranking, low-confidence retraction, and all-option evidence calibration.
92.89% validation accuracy and 92.63% official test accuracy.
Lai
Track 1
Cold-start structured reasoning data, elimination-guided reward design, and option-perturbed DAPO that rewards semantic candidate identity rather than displayed option letters.
92.75% validation accuracy and 92.46% official challenge accuracy.
MediaLab
Track 2
Scenario-aware service agent, official-tool canonicalization, evidence ledger, and scope-limited correction for state-changing calls and final replies.
0.844 Joint and 0.916 Micro over five test scenarios.
GML-MM Group
Track 2
Status-aware grounded execution, Reporter summaries for evidence compression, step-specific tool views, and paired visual grounding for occluded references.
0.781 Joint on released retail1 ; 0.823 Joint over five test scenarios.
Table 4. Representative participant teams and their main technical mechanisms.