MARS converts long videos to captions and summaries, maintains modality-specific memories, and deploys an agent to select evidence or answer, placing second on the CASTLE Challenge leaderboard.
Attribute-guided collab- orative learning for partial person re-identification.IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(12):14144–14160, 2023
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MARS: Technical Report for the CASTLE Challenge at EgoVis 2026
MARS converts long videos to captions and summaries, maintains modality-specific memories, and deploys an agent to select evidence or answer, placing second on the CASTLE Challenge leaderboard.