MedVidU @ ECCV 2026 is a workshop focused on advancing video understanding for surgical and clinical procedures, bringing together researchers in computer vision and healthcare. It addresses challenges in interpreting medical video data through tasks like surgical phase recognition, tool tracking, and dense captioning, supported by the MedVidBench dataset across four domains: laparoscopic, open, robotic surgery, and nursing.
A strong submission should clearly identify its contribution and evaluate it appropriately.
Workshop Paper
Submissions reporting novel theories, methods, and applications of video understanding in medical and surgical settings, formatted according to ECCV 2026 author guidelines and intended for publication in the ECCV 2026 workshop proceedings.
Challenge Paper
Mandatory submission for MedVidU Challenge participants, reporting methods used in the challenge, submitted via OpenReview alongside challenge participation.
Research areas in scope
01
Topics of Interest
Temporal action grounding in medical videoDense video captioning for medical proceduresSurgical tool detection and trackingMedical video foundation models and multimodal large language modelsSurgical phase and action recognitionMedical scene understanding from videos and 3D/4D reconstructionPrivacy-preserving analysis of surgical recordingsExplainability and trustworthiness in AI for medical videosEfficient video representation learning for long medical procedures
Policies worth checking twice
Submissions must be formatted according to ECCV 2026 author guidelines.
Accepted papers are intended to be published in the ECCV 2026 workshop proceedings.
MedVidU Challenge participants are required to submit a paper.
You can submit a paper without participating in the MedVidU Challenge.
All deadlines are Anywhere on Earth (AoE).
Challenge participants must submit model name, team name, authors, institution, email ID, and set Remark: "ECCV challenge".
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.