File size: 3,612 Bytes
17c6d62 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 |
<!--Copyright 2022 The HuggingFace Team. All rights reserved.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with
the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on
an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the
specific language governing permissions and limitations under the License.
β οΈ Note that this file is in Markdown but contain specific syntax for our doc-builder (similar to MDX) that may not be
rendered properly in your Markdown viewer.
-->
# TimeSformer [[timesformer]]
## κ°μ [[overview]]
TimeSformer λͺ¨λΈμ Facebook Researchμμ μ μν [TimeSformer: Is Space-Time Attention All You Need for Video Understanding?](https://arxiv.org/abs/2102.05095)μμ μκ°λμμ΅λλ€. μ΄ μ°κ΅¬λ 첫 λ²μ§Έ λΉλμ€ Transformerλ‘μ, νλ μΈμ λΆμΌμμ μ€μν μ΄μ νκ° λμμ΅λλ€. λν Transformer κΈ°λ°μ λΉλμ€ μ΄ν΄ λ° λΆλ₯ λ
Όλ¬Έμ λ§μ μκ°μ μ£Όμμ΅λλ€.
λ
Όλ¬Έμ μ΄λ‘μ λ€μκ³Ό κ°μ΅λλ€.
*μ°λ¦¬λ 곡κ°κ³Ό μκ°μ κ±Έμ³ μ
ν μ΄ν
μ
λ§μ μ¬μ©νλ ν©μ±κ³±μ΄ μλ(convolution-free) λΉλμ€ λΆλ₯ λ°©λ²μ μ μν©λλ€. μ΄ λ°©λ²μ βTimeSformerβλΌκ³ λΆλ¦¬λ©°, νμ€ Transformer μν€ν
μ²λ₯Ό λΉλμ€μ μ μ©νμ¬ νλ μ μμ€ ν¨μΉ μνμ€λ‘λΆν° μ§μ μ곡κ°μ νΉμ§μ νμ΅ν μ μκ² ν©λλ€. μ°λ¦¬μ μ€νμ μ°κ΅¬λ λ€μν μ
ν μ΄ν
μ
λ°©μμ λΉκ΅νλ©°, μκ°μ μ΄ν
μ
κ³Ό 곡κ°μ μ΄ν
μ
μ κ°κ°μ λΈλ‘ λ΄μμ λ³λλ‘ μ μ©νλ βλΆν μ΄ν
μ
β λ°©μμ΄ κ³ λ €λ μ€κ³ μ ν μ€ κ°μ₯ μ°μν λΉλμ€ λΆλ₯ μ νλλ₯Ό μ 곡νλ€λ κ²μ μμ¬ν©λλ€. μ΄ νμ μ μΈ μ€κ³μλ λΆκ΅¬νκ³ , TimeSformerλ Kinetics-400 λ° Kinetics-600μ ν¬ν¨ν μ¬λ¬ νλ μΈμ λ²€μΉλ§ν¬μμ μ΅μ²¨λ¨ κ²°κ³Όλ₯Ό λ¬μ±νμΌλ©°, νμ¬κΉμ§ λ³΄κ³ λ κ°μ₯ λμ μ νλλ₯Ό κΈ°λ‘νμ΅λλ€. λ§μ§λ§μΌλ‘, 3D ν©μ±κ³± λ€νΈμν¬μ λΉκ΅νμ λ, TimeSformerλ λ λΉ λ₯΄κ² νμ΅ν μ μμΌλ©°, μ½κ°μ μ νλ μ νλ₯Ό κ°μνλ©΄ ν
μ€νΈ ν¨μ¨μ±μ΄ ν¬κ² ν₯μλκ³ , 1λΆ μ΄μμ κΈ΄ λΉλμ€ ν΄λ¦½μλ μ μ©ν μ μμ΅λλ€. μ½λμ λͺ¨λΈμ λ€μ λ§ν¬μμ νμΈν μ μμ΅λλ€: [https URL λ§ν¬](https://github.com/facebookresearch/TimeSformer).*
μ΄ λͺ¨λΈμ [fcakyon](https://huggingface.co/fcakyon)μ΄ κΈ°μ¬νμμ΅λλ€.
μλ³Έ μ½λλ [μ¬κΈ°](https://github.com/facebookresearch/TimeSformer)μμ νμΈν μ μμ΅λλ€.
## μ¬μ© ν [[usage-tips]]
λ€μν μ¬μ νμ΅λ λͺ¨λΈμ λ³νλ€μ΄ μμ΅λλ€. μ¬μ©νλ €λ λ°μ΄ν°μ
μ λ§μΆ° μ¬μ νμ΅λ λͺ¨λΈμ μ νν΄μΌ ν©λλ€. λν, λͺ¨λΈ ν¬κΈ°μ λ°λΌ ν΄λ¦½λΉ μ
λ ₯ νλ μ μκ° λ¬λΌμ§λ―λ‘, μ¬μ νμ΅λ λͺ¨λΈμ μ νν λ μ΄ λ§€κ°λ³μλ₯Ό κ³ λ €ν΄μΌ ν©λλ€.
## 리μμ€ [[resources]]
- [Video classification task guide](../tasks/video_classification)
## TimesformerConfig [[transformers.TimesformerConfig]]
[[autodoc]] TimesformerConfig
## TimesformerModel [[transformers.TimesformerModel]]
[[autodoc]] TimesformerModel
- forward
## TimesformerForVideoClassification [[transformers.TimesformerForVideoClassification]]
[[autodoc]] TimesformerForVideoClassification
- forward |