RemoteTrimmer: Adaptive Structural Pruning for Remote Sensing Image Classification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zou, Guangwenjie, Yao, Liang, Liu, Fan, Zhang, Chuanyi, Li, Xin, Chen, Ning, Xu, Shengxiang, Zhou, Jun
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917873435279360
author Zou, Guangwenjie
Yao, Liang
Liu, Fan
Zhang, Chuanyi
Li, Xin
Chen, Ning
Xu, Shengxiang
Zhou, Jun
author_facet Zou, Guangwenjie
Yao, Liang
Liu, Fan
Zhang, Chuanyi
Li, Xin
Chen, Ning
Xu, Shengxiang
Zhou, Jun
contents Since high resolution remote sensing image classification often requires a relatively high computation complexity, lightweight models tend to be practical and efficient. Model pruning is an effective method for model compression. However, existing methods rarely take into account the specificity of remote sensing images, resulting in significant accuracy loss after pruning. To this end, we propose an effective structural pruning approach for remote sensing image classification. Specifically, a pruning strategy that amplifies the differences in channel importance of the model is introduced. Then an adaptive mining loss function is designed for the fine-tuning process of the pruned model. Finally, we conducted experiments on two remote sensing classification datasets. The experimental results demonstrate that our method achieves minimal accuracy loss after compressing remote sensing classification models, achieving state-of-the-art (SoTA) performance.
format Preprint
id arxiv_https___arxiv_org_abs_2412_12603
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle RemoteTrimmer: Adaptive Structural Pruning for Remote Sensing Image Classification
Zou, Guangwenjie
Yao, Liang
Liu, Fan
Zhang, Chuanyi
Li, Xin
Chen, Ning
Xu, Shengxiang
Zhou, Jun
Computer Vision and Pattern Recognition
Since high resolution remote sensing image classification often requires a relatively high computation complexity, lightweight models tend to be practical and efficient. Model pruning is an effective method for model compression. However, existing methods rarely take into account the specificity of remote sensing images, resulting in significant accuracy loss after pruning. To this end, we propose an effective structural pruning approach for remote sensing image classification. Specifically, a pruning strategy that amplifies the differences in channel importance of the model is introduced. Then an adaptive mining loss function is designed for the fine-tuning process of the pruned model. Finally, we conducted experiments on two remote sensing classification datasets. The experimental results demonstrate that our method achieves minimal accuracy loss after compressing remote sensing classification models, achieving state-of-the-art (SoTA) performance.
title RemoteTrimmer: Adaptive Structural Pruning for Remote Sensing Image Classification
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.12603