2st CASTLE Grand Challenge

The second CASTLE Challenge will be held at ACM Multimedia 2026 in Rio de Janeiro, Brazil.

To submit your paper and your results (for participants of the automatic track), please use the submission form on OpenReview

Timeline

Tasks

The CASTLE Challenge features a diverse set of tasks, including event detection, retrieval, and question answering. Future editions will expand the scope, but for this edition, the tasks include:

Given a textual description (in English), participants must identify all timeframes where a specific event occurs. Events should be reported with both a time range and a video ID.

Given a textual (in English) or visual (i.e., using an image) example of a physical object, participants must find all occurrences of that object across any of the video streams.

๐Ÿ’ฌ Question Answering

Given a question in natural language (in English), participants must provide an answer. The response should be formulated in natural language and include references to relevant sensor streams and time intervals as supporting evidence.


Evaluation

The challenge will operate across two tracks: fully-automatic and interactive.

โš™๏ธ Fully-Automatic Track

Participants receive queries in advance and generate results using any method they choose. These results are then submitted to the challenge organizers for evaluation. Please see the list of queries below.

๐ŸŽฎ Interactive Track

This track will be evaluated live during the conference. Participants must solve tasks synchronously and interactively within a limited timeframe. This format follows established competitions such as the Video Browser Showdown and the Lifelog Search Challenge.


Queries

๐Ÿ” Event Instance Search

๐Ÿ“ฆ Object Instance Search

๐Ÿ’ฌ Question Answering


Submission Format

โš™๏ธ Fully-Automatic Track

For the fully-automatic track, please submit your results together with your paper on OpenReview. Results should be submitted in CSV format with one file per task type (i.e., three files, one each for the ๐Ÿ” Event Instance Search, ๐Ÿ“ฆ Object Instance Search, and ๐Ÿ’ฌ Question Answering tasks) in a ZIP file. The name of each file should indicate the task type. The columns for the files are listed below.

๐Ÿ” Event Instance Search

๐Ÿ“ฆ Object Instance Search

๐Ÿ’ฌ Question Answering

๐ŸŽฎ Interactive Track

The interactive track will be evaluated during a hybrid session at ACM Multimedia 2026. Systems are supposed to submit their task solutions to the Distributed Retrieval Evaluation Server via its API.