# ETAD: Training Action Detection End to End on a Laptop

arXiv: https://arxiv.org/abs/2205.07134 · HTML: https://mlsanity.com/paper/2205.07134/

## Key facts

- Cited by 10 papers in the map (data through 14 Oct 2026).
- Builds on 37 of its 58 references that are in the map.
- First posted on arXiv 14 May 2022.

## Builds on (most cited first, 30 of 37)

- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [The Kinetics Human Action Video Dataset](https://mlsanity.com/paper/1705.06950/) (2017)
- [SlowFast Networks for Video Recognition](https://mlsanity.com/paper/1812.03982/) (2018)
- [Cascade R-CNN: Delving into High Quality Object Detection](https://mlsanity.com/paper/1712.00726/) (2017)
- [Mixed Precision Training](https://mlsanity.com/paper/1710.03740/) (2017)
- [A Closer Look at Spatiotemporal Convolutions for Action Recognition](https://mlsanity.com/paper/1711.11248/) (2017)
- [TSM: Temporal Shift Module for Efficient Video Understanding](https://mlsanity.com/paper/1811.08383/) (2018)
- [Training Deep Nets with Sublinear Memory Cost](https://mlsanity.com/paper/1604.06174/) (2016)
- [BMN: Boundary-Matching Network for Temporal Action Proposal Generation](https://mlsanity.com/paper/1907.09702/) (2019)
- [ActionFormer: Localizing Moments of Actions with Transformers](https://mlsanity.com/paper/2202.07925/) (2022)
- [BSN: Boundary Sensitive Network for Temporal Action Proposal Generation](https://mlsanity.com/paper/1806.02964/) (2018)
- [Temporal Action Detection with Structured Segment Networks](https://mlsanity.com/paper/1704.06228/) (2017)
- [Libra R-CNN: Towards Balanced Learning for Object Detection](https://mlsanity.com/paper/1904.02701/) (2019)
- [G-TAD: Sub-Graph Localization for Temporal Action Detection](https://mlsanity.com/paper/1911.11462/) (2019)
- [Graph Convolutional Networks for Temporal Action Localization](https://mlsanity.com/paper/1909.03252/) (2019)
- [Rethinking the Faster R-CNN Architecture for Temporal Action Localization](https://mlsanity.com/paper/1804.07667/) (2018)
- [R-C3D: Region Convolutional 3D Network for Temporal Activity Detection](https://mlsanity.com/paper/1703.07814/) (2017)
- [Learning Salient Boundary Feature for Anchor-free Temporal Action Localization](https://mlsanity.com/paper/2103.13137/) (2021)
- [SCSampler: Sampling Salient Clips from Video for Efficient Action Recognition](https://mlsanity.com/paper/1904.04289/) (2019)
- [UntrimmedNets for Weakly Supervised Action Recognition and Detection](https://mlsanity.com/paper/1703.03329/) (2017)
- [Relaxed Transformer Decoders for Direct Action Proposal Generation](https://mlsanity.com/paper/2102.01894/) (2021)
- [Temporal Context Aggregation Network for Temporal Action Proposal Refinement](https://mlsanity.com/paper/2103.13141/) (2021)
- [Video Self-Stitching Graph Network for Temporal Action Localization](https://mlsanity.com/paper/2011.14598/) (2020)
- [Fast Learning of Temporal Action Proposal via Dense Boundary Generator](https://mlsanity.com/paper/1911.04127/) (2019)
- [Boundary Content Graph Neural Network for Temporal Action Proposal Generation](https://mlsanity.com/paper/2008.01432/) (2020)
- [TALLFormer: Temporal Action Localization with a Long-memory Transformer](https://mlsanity.com/paper/2204.01680/) (2022)
- [Multi-granularity Generator for Temporal Action Proposal](https://mlsanity.com/paper/1811.11524/) (2018)
- [TSP: Temporally-Sensitive Pretraining of Video Encoders for Localization Tasks](https://mlsanity.com/paper/2011.11479/) (2020)
- [CUHK & ETHZ & SIAT Submission to ActivityNet Challenge 2016](https://mlsanity.com/paper/1608.00797/) (2016)

## Built on by (most cited first, 10 of 10)

- [End-to-End Temporal Action Detection with 1B Parameters Across 1000 Frames](https://mlsanity.com/paper/2311.17241/) (2023)
- [Action Sensitivity Learning for Temporal Action Localization](https://mlsanity.com/paper/2305.15701/) (2023)
- [Temporal Action Localization with Enhanced Instant Discriminability](https://mlsanity.com/paper/2309.05590/) (2023)
- [Harnessing Temporal Causality for Advanced Temporal Action Detection](https://mlsanity.com/paper/2407.17792/) (2024)
- [TimeLoc: A Unified End-to-End Framework for Precise Timestamp Localization in Long Videos](https://mlsanity.com/paper/2503.06526/) (2025)
- [A benchmark for video-based laparoscopic skill analysis and assessment](https://mlsanity.com/paper/2602.09927/) (2026)
- [FDDet: Frequency-Decoupling for Boundary Refinement in Temporal Action Detection](https://mlsanity.com/paper/2504.00647/) (2025)
- [A Paradigm Shift: Fully End-to-End Training for Temporal Sentence Grounding in Videos](https://mlsanity.com/paper/2604.02860/) (2026)
- [Efficient Spatial-Temporal Focal Adapter with SSM for Temporal Action Detection](https://mlsanity.com/paper/2604.09164/) (2026)
- [Decoupling Spatio-Temporal Adapter for Fine-Grained Badminton Action Localization](https://mlsanity.com/paper/2605.23355/) (2026)

## Abstract (arXiv)

Temporal action detection (TAD) with end-to-end training often suffers from the pain of huge demand for computing resources due to long video duration. In this work, we propose an efficient temporal action detector (ETAD) that can train directly from video frames with extremely low GPU memory consumption. Our main idea is to minimize and balance the heavy computation among features and gradients in each training iteration. We propose to sequentially forward the snippet frame through the video encoder, and backward only a small necessary portion of gradients to update the encoder. To further alleviate the computational redundancy in training, we propose to dynamically sample only a small subset of proposals during training. Moreover, various sampling strategies and ratios are studied for both the encoder and detector. ETAD achieves state-of-the-art performance on TAD benchmarks with remarkable efficiency. On ActivityNet-1.3, training ETAD in 18 hours can reach 38.25% average mAP with only 1.3 GB memory consumption per video under end-to-end training. Our code will be publicly released.