computer-vision-11

Feature Compression

3260 papers • 126 benchmarks • 313 datasets

Compress data for machine interpretability to perform downstream tasks, rather than for human perception.

(Image credit: Papersgraph)

Benchmarks

These leaderboards are used to track progress in feature-compression-11

Trend

Dataset

Best Model

Actions

No benchmarks available.

Libraries

i

Use these libraries to find feature-compression-11 models and implementations

Datasets

No datasets available.

Subtasks

No subtasks available.

Most implemented papers

Supervised Compression for Resource-Constrained Edge Computing Systems

M. Levorato, Yoshitomo Matsubara, Ruihan Yang, Stephan Mandt•Fri Aug 20 2021

This paper adopts ideas from knowledge distillation and neural image compression to compress intermediate feature representations more efficiently and shows that the learned feature representations can be tuned to serve multiple downstream tasks.

71

Content

0

Paper Graph

VIMI: Vehicle-Infrastructure Multi-view Intermediate Fusion for Camera-based 3D Object Detection

Siqi Fan, Zhe Wang, Xiao Huo, Tongda Xu, Yan Wang, Jingjing Liu, Yilun Chen, Ya-Qin Zhang•Sun Mar 19 2023

This work proposes a novel 3D object detection framework, Vehicles-Infrastructure Multi-view Intermediate fusion (VIMI), and proposes a Multi-scale Cross Attention (MCA) module that fuses infrastructure and vehicle features on selective multi-scales to correct the calibration noise introduced by camera asynchrony.

26 0

Paper Graph

Flexible Variable-Rate Image Feature Compression for Edge-Cloud Systems

Md Adnan Faisal Hossain, Zhihao Duan, Yuning Huang, F. Zhu•Fri Jun 30 2023

A flexible variable-rate feature compression method is presented that can operate on a range of rates by introducing a rate control parameter as an input to the neural network model.

7 0

Paper Graph

Context-Aware Deep Feature Compression for High-Speed Visual Tracking

Sangdoo Yun, Jongwon Choi, J. Choi, Y. Demiris, H. Chang, Tobias Fischer, Jiyeoup Jeong, Kyuewang Lee•Tue Mar 27 2018

A new context-aware correlation filter based tracking framework to achieve both high computational speed and state-of-the-art performance among real-time trackers and introduces extrinsic denoising processes and a new orthogonality loss term for pre-training and fine-tuning of the expert autoencoders.

198 0

Paper Graph

BottleNet++: An End-to-End Approach for Feature Compression in Device-Edge Co-Inference Systems

Jiawei Shao, Jun Zhang•Wed Oct 30 2019

An end-to-end architecture that consists of an encoder, a non-trainable channel layer, and a decoder for more efficient feature compression and transmission, which achieves a much higher compression ratio than existing methods.

199 0

Paper Graph

Lossy Compression for Lossless Prediction

Chris J. Maddison, Yann Dubois, Benjamin Bloem-Reddy•Sun Jun 20 2021

This paper describes the bit-rate required to ensure high performance on all predictive tasks that are invariant under a set of transformations, such as data augmentations, and designs unsupervised objectives for training neural compressors.

75 0

Paper Graph

Context-Aware Compilation of DNN Training Pipelines across Edge and Cloud

C. Li, Dixi Yao, Liyao Xiang, Zifan Wang, Jiayu Xu, Xinbing Wang•Sun Dec 26 2021

Experimental results show that the proposed pipeline training framework not only significantly speeds up training, but also incurs little accuracy loss or additional memory/energy overhead, delivering a practical and efficient solution to edge-cloud model training.

22 0

Paper Graph

SC2 Benchmark: Supervised Compression for Split Computing

M. Levorato, Yoshitomo Matsubara, Ruihan Yang, Stephan Mandt•Tue Mar 15 2022

This study introduces supervised compression for split computing (SC2) and proposes new evaluation criteria: minimizing computation on the mobile device, minimizing transmitted data size, and maximizing model accuracy and releases sc2bench, a Python package for future research on SC2.

26 0

Paper Graph

Multi-Agent Collaborative Inference via DNN Decoupling: Intermediate Feature Compression and Edge Learning

Shiwen Mao, Z. Hao, Guanyu Xu, Yong Luo, Han Hu, Jianping An•Mon May 23 2022

A lightweight autoencoder-based method to compress the large intermediate feature of DNN, then a multi-agent hybrid proximal policy optimization (MAHPPO) algorithm to solve the optimization problem with a hybrid action space is proposed.

46 0

Paper Graph

Compressing Features for Learning With Noisy Labels

S. Hu, Yingyi Chen, Xin Shen, C. Ai, J. Suykens•Sun Jun 26 2022

This article revisits one classical regularization named Dropout and its variant Nested Dropout and revisits the problem of learning with noisy labels and introduces compression inductive bias to network architectures to alleviate this overfitting problem.

23 0

Paper Graph

Adding a benchmark result helps the community track progress.