Tiny Video Networks
Abstract Automatic video understanding is becoming more important for applications where real‐time performance is crucial and compute is limited: for example, automated video tagging, robot perception, activity recognition for mobile devices. Yet, accurate solutions so far have been computationally...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2022-02-01
|
| Series: | Applied AI Letters |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/ail2.38 |
| _version_ | 1819018430700847104 |
|---|---|
| author | A. J. Piergiovanni Anelia Angelova Michael S. Ryoo |
| author_facet | A. J. Piergiovanni Anelia Angelova Michael S. Ryoo |
| author_sort | A. J. Piergiovanni |
| collection | DOAJ |
| description | Abstract Automatic video understanding is becoming more important for applications where real‐time performance is crucial and compute is limited: for example, automated video tagging, robot perception, activity recognition for mobile devices. Yet, accurate solutions so far have been computationally intensive. We propose efficient models for videos—Tiny Video Networks—which are video architectures, automatically designed to comply with fast runtimes and, at the same time are effective at video recognition tasks. The TVNs run at faster‐than‐real‐time speeds and demonstrate strong performance across several video benchmarks. These models not only provide new tools for real‐time video applications, but also enable fast research and development in video understanding. Code and models are available. |
| first_indexed | 2024-12-21T03:19:18Z |
| format | Article |
| id | doaj.art-ddf5c0cc45f94ec78f1505e61ceda574 |
| institution | Directory Open Access Journal |
| issn | 2689-5595 |
| language | English |
| last_indexed | 2024-12-21T03:19:18Z |
| publishDate | 2022-02-01 |
| publisher | Wiley |
| record_format | Article |
| series | Applied AI Letters |
| spelling | doaj.art-ddf5c0cc45f94ec78f1505e61ceda5742022-12-21T19:17:44ZengWileyApplied AI Letters2689-55952022-02-0131n/an/a10.1002/ail2.38Tiny Video NetworksA. J. Piergiovanni0Anelia Angelova1Michael S. Ryoo2Google Research, Robotics at Google Mountain View California USAGoogle Research, Robotics at Google Mountain View California USAGoogle Research, Robotics at Google Mountain View California USAAbstract Automatic video understanding is becoming more important for applications where real‐time performance is crucial and compute is limited: for example, automated video tagging, robot perception, activity recognition for mobile devices. Yet, accurate solutions so far have been computationally intensive. We propose efficient models for videos—Tiny Video Networks—which are video architectures, automatically designed to comply with fast runtimes and, at the same time are effective at video recognition tasks. The TVNs run at faster‐than‐real‐time speeds and demonstrate strong performance across several video benchmarks. These models not only provide new tools for real‐time video applications, but also enable fast research and development in video understanding. Code and models are available.https://doi.org/10.1002/ail2.38efficient video modelsvideo architecture searchvideo understanding |
| spellingShingle | A. J. Piergiovanni Anelia Angelova Michael S. Ryoo Tiny Video Networks Applied AI Letters efficient video models video architecture search video understanding |
| title | Tiny Video Networks |
| title_full | Tiny Video Networks |
| title_fullStr | Tiny Video Networks |
| title_full_unstemmed | Tiny Video Networks |
| title_short | Tiny Video Networks |
| title_sort | tiny video networks |
| topic | efficient video models video architecture search video understanding |
| url | https://doi.org/10.1002/ail2.38 |
| work_keys_str_mv | AT ajpiergiovanni tinyvideonetworks AT aneliaangelova tinyvideonetworks AT michaelsryoo tinyvideonetworks |