SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.16559 · ICCV · 2021

Broaden Your Views for Self-Supervised Video Learning

Andrew Zisserman, Michal Valko, Jean-Baptiste Alayrac, Adrià Recasens, Jean-Bastien Grill, Florent Altché, Florian Strub, Aäron Van Den Oord, Mateusz Malinowski, Pauline Luc, Corentin Tallec, Viorica Pȃtrȃucean, and 1 more

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 8 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
deepmind/brave canonical 7 of 8
FunctionStatusWhere it lives
compute_linearly_spaced_sample_indices Ran deepmind/brave/brave/datasets/sampling.py
code served (permissive licence) · get_code("d72aee9f7872a1f4")
extend_sequence Ran deepmind/brave/brave/datasets/media_sequences.py
code served (permissive licence) · get_code("4e1f5f77cae09400")
md5 Ran deepmind/brave/brave/download_hmdb.py
code served (permissive licence) · get_code("8bd26bff58d47d3a")
pcm_to_log_mel_spectrogram Ran deepmind/brave/brave/datasets/spectrograms.py
code served (permissive licence) · get_code("f7f43698d451909c")
random_sample Ran deepmind/brave/brave/datasets/sampling.py
code served (permissive licence) · get_code("e6ad315cfb7d7612")
tf_record_shard_reader Ran deepmind/brave/brave/datasets/media_sequences.py
code served (permissive licence) · get_code("ca07ab87c895635c")
write_tf_record_dataset_fixture Ran deepmind/brave/brave/datasets/fixtures.py
code served (permissive licence) · get_code("4c090fcd18780345")
media_sequence_dataset Not yet run deepmind/brave/brave/datasets/media_sequences.py
code served (permissive licence) · get_code("315e821fb79e66f7")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Most successful self-supervised learning methods are trained to align the representations of two independent views from the data. State-of-the-art methods in video are inspired by image techniques, where these two views are similarly extracted by cropping and augmenting the resulting crop. However, these methods miss a crucial element in the video domain: time. We introduce BraVe, a self-supervised learning framework for video. In BraVe, one of the views has access to a narrow temporal window of the video while the other view has a broad access to the video content. Our models learn to generalise from the narrow view to the general content of the video. Furthermore, BraVe processes the views with different backbones, enabling the use of alternative augmentations or modalities into the broad view such as optical flow, randomly convolved RGB frames, audio or their combinations. We demonstrate that BraVe achieves stateof-the-art results in self-supervised representation learning on standard video and audio classification benchmarks including UCF101, HMDB51, Kinetics, ESC-50 and AudioSet.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2103.16559")
get_code_for_paper("2103.16559")
have("2103.16559")

Connect an agent — have() is free.