# ELASTIC: Improving CNNs with Dynamic Scaling Policies

arXiv: https://arxiv.org/abs/1812.05262 · HTML: https://mlsanity.com/paper/1812.05262/

## Key facts

- Cited by 20 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Dec 2018.

## Built on by (most cited first, 20 of 20)

- [Dynamic Neural Networks: A Survey](https://mlsanity.com/paper/2102.04906/) (2021)
- [GhostNetV2: Enhance Cheap Operation with Long-Range Attention](https://mlsanity.com/paper/2211.12905/) (2022)
- [Dynamic Resolution Network](https://mlsanity.com/paper/2106.02898/) (2021)
- [Dual Vision Transformer](https://mlsanity.com/paper/2207.04976/) (2022)
- [UDIS: Unsupervised Discovery of Bias in Deep Visual Recognition Models](https://mlsanity.com/paper/2110.15499/) (2021)
- [SHARP: Environment and Person Independent Activity Recognition with Commodity IEEE 802.11 Access Points](https://mlsanity.com/paper/2103.09924/) (2021)
- [Hard-Attention for Scalable Image Classification](https://mlsanity.com/paper/2102.10212/) (2021)
- [Task-Customized Self-Supervised Pre-training with Scalable Dynamic Routing](https://mlsanity.com/paper/2205.13267/) (2022)
- [Scale-invariant scale-channel networks: Deep networks that generalise to previously unseen scales](https://mlsanity.com/paper/2106.06418/) (2021)
- [Boosted Dynamic Neural Networks](https://mlsanity.com/paper/2211.16726/) (2022)
- [AsConvSR: Fast and Lightweight Super-Resolution Network with Assembled Convolutions](https://mlsanity.com/paper/2305.03387/) (2023)
- [Depth Super-Resolution from Explicit and Implicit High-Frequency Features](https://mlsanity.com/paper/2303.09307/) (2023)
- [AesFA: An Aesthetic Feature-Aware Arbitrary Neural Style Transfer](https://mlsanity.com/paper/2312.05928/) (2023)
- [Dynamic Low-Resolution Distillation for Cost-Efficient End-to-End Text Spotting](https://mlsanity.com/paper/2207.06694/) (2022)
- [DyRA: Portable Dynamic Resolution Adjustment Network for Existing Detectors](https://mlsanity.com/paper/2311.17098/) (2023)
- [EdgeSync: Faster Edge-model Updating via Adaptive Continuous Learning for Video Data Drift](https://mlsanity.com/paper/2406.03001/) (2024)
- [MR-RawNet: Speaker verification system with multiple temporal resolutions for variable duration utterances using raw waveforms](https://mlsanity.com/paper/2406.07103/) (2024)
- [Temporally Resolution Decrement: Utilizing the Shape Consistency for Higher Computational Efficiency](https://mlsanity.com/paper/2112.00954/) (2021)
- [Mondrian: On-Device High-Performance Video Analytics with Compressive Packed Inference](https://mlsanity.com/paper/2403.07598/) (2024)
- [EdgeSync: Accelerating Edge-Model Updates for Data Drift through Adaptive Continuous Learning](https://mlsanity.com/paper/2510.21781/) (2025)

## Abstract (arXiv)

Scale variation has been a challenge from traditional to modern approaches in computer vision. Most solutions to scale issues have a similar theme: a set of intuitive and manually designed policies that are generic and fixed (e.g. SIFT or feature pyramid). We argue that the scaling policy should be learned from data. In this paper, we introduce ELASTIC, a simple, efficient and yet very effective approach to learn a dynamic scale policy from data. We formulate the scaling policy as a non-linear function inside the network's structure that (a) is learned from data, (b) is instance specific, (c) does not add extra computation, and (d) can be applied on any network architecture. We applied ELASTIC to several state-of-the-art network architectures and showed consistent improvement without extra (sometimes even lower) computation on ImageNet classification, MSCOCO multi-label classification, and PASCAL VOC semantic segmentation. Our results show major improvement for images with scale challenges. Our code is available here: https://github.com/allenai/elastic