ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models.
Ilker KesenAndrea PedrottiMustafa DoganMichele CafagnaEmre Can AcikgozLetitia ParcalabescuIacer CalixtoAnette FrankAlbert GattAykut ErdemErkut ErdemPublished in: CoRR (2023)
Keyphrases
- language model
- language modeling
- temporal information
- n gram
- retrieval model
- information retrieval
- probabilistic model
- video sequences
- language modelling
- statistical language models
- query expansion
- video data
- document retrieval
- test collection
- video content
- speech recognition
- smoothing methods
- multimedia
- context sensitive
- video frames
- language models for information retrieval
- vector space model
- ad hoc information retrieval
- natural language processing
- natural language
- language model for information retrieval
- pseudo relevance feedback
- relevance model
- video search
- video retrieval
- video shots
- translation model