1

Frozen in Time: A Joint Video and Image Encoder for End-to-End Retrieval [ICCV'21]
Text-to-video retrieval using a dual encoder architecture of text and video transformers, which trains and test on videos and images jointly.
Max Bain, Arsha Nagrani, Gül Varol, Andrew Zisserman
PDF Cite Code Dataset Project Demo
Frozen in Time: A Joint Video and Image Encoder for End-to-End Retrieval [ICCV'21]
Condensed Movies: Story Based Retrieval with Contextual Embeddings [ACCV'20]
Retrieving short movie clips using a mixture of experts model with movie-level context. Includes the release of a large 34k+ clips, 3k+ movies dataset with captions.
Max Bain, Arsha Nagrani, Andrew Brown, Andrew Zisserman
PDF Cite Code Dataset Challenge
Condensed Movies: Story Based Retrieval with Contextual Embeddings [ACCV'20]
Count, Crop and Recognise: Fine-Grained Recognition in the Wild [ICCVW'19]
Fine-grained recognition of chimpanzee identities via full-frame.
Max Bain, Arsha Nagrani, Daniel Schofield, Andrew Zisserman
PDF Cite Project
Count, Crop and Recognise: Fine-Grained Recognition in the Wild [ICCVW'19]