DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation

The accuracy of neural networks heavily depends on numerous precisely annotated samples. But in actual datasets, the number of samples for each category varies greatly. In the dataset, some classes may have a large sample size, called majority classes, while others may have a small sample size, call...

Full description

Saved in:
Bibliographic Details
Main Authors: Xiaohong Liu, Lin Wang, Lijing Ma, Chaoli Wang
Format: Article
Language:English
Published: Taylor & Francis Group 2024-12-01
Series:Applied Artificial Intelligence
Online Access:https://www.tandfonline.com/doi/10.1080/08839514.2024.2411845
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846119956577517568
author Xiaohong Liu
Lin Wang
Lijing Ma
Chaoli Wang
author_facet Xiaohong Liu
Lin Wang
Lijing Ma
Chaoli Wang
author_sort Xiaohong Liu
collection DOAJ
description The accuracy of neural networks heavily depends on numerous precisely annotated samples. But in actual datasets, the number of samples for each category varies greatly. In the dataset, some classes may have a large sample size, called majority classes, while others may have a small sample size, called minority classes. Training model with imbalanced data is often conducive to the majority classes, while unfair to minority classes. As a result, the outputs return good performance on majority classes and bad performance on minority classes. This article proposes a new loss function called dynamic-recall focal loss (DRFL), which can solve the problem of imbalanced data categories in image classification and medical segmentation tasks. The DRFL assigns different weight coefficient to the classes according to their dynamic recall based on focal loss. Experimental results have shown that the newly proposed loss function DRFL can effectively improve the classification and segmentation accuracy of two imbalanced datasets.
format Article
id doaj-art-db73953c903a40c09725b264ac6b55ef
institution Kabale University
issn 0883-9514
1087-6545
language English
publishDate 2024-12-01
publisher Taylor & Francis Group
record_format Article
series Applied Artificial Intelligence
spelling doaj-art-db73953c903a40c09725b264ac6b55ef2024-12-16T16:13:01ZengTaylor & Francis GroupApplied Artificial Intelligence0883-95141087-65452024-12-0138110.1080/08839514.2024.2411845DRFL: Dynamic-Recall Focal Loss for Image Classification and SegmentationXiaohong Liu0Lin Wang1Lijing Ma2Chaoli Wang3Department of Radiology, Shanghai Eighth People’s Hospital, Shanghai, ChinaDepartment of Computer Science, East China Normal, Shanghai, ChinaDepartment of Control Science and Engineering, University of Shanghai for Science and Technology, Shanghai, ChinaDepartment of Control Science and Engineering, University of Shanghai for Science and Technology, Shanghai, ChinaThe accuracy of neural networks heavily depends on numerous precisely annotated samples. But in actual datasets, the number of samples for each category varies greatly. In the dataset, some classes may have a large sample size, called majority classes, while others may have a small sample size, called minority classes. Training model with imbalanced data is often conducive to the majority classes, while unfair to minority classes. As a result, the outputs return good performance on majority classes and bad performance on minority classes. This article proposes a new loss function called dynamic-recall focal loss (DRFL), which can solve the problem of imbalanced data categories in image classification and medical segmentation tasks. The DRFL assigns different weight coefficient to the classes according to their dynamic recall based on focal loss. Experimental results have shown that the newly proposed loss function DRFL can effectively improve the classification and segmentation accuracy of two imbalanced datasets.https://www.tandfonline.com/doi/10.1080/08839514.2024.2411845
spellingShingle Xiaohong Liu
Lin Wang
Lijing Ma
Chaoli Wang
DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
Applied Artificial Intelligence
title DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
title_full DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
title_fullStr DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
title_full_unstemmed DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
title_short DRFL: Dynamic-Recall Focal Loss for Image Classification and Segmentation
title_sort drfl dynamic recall focal loss for image classification and segmentation
url https://www.tandfonline.com/doi/10.1080/08839514.2024.2411845
work_keys_str_mv AT xiaohongliu drfldynamicrecallfocallossforimageclassificationandsegmentation
AT linwang drfldynamicrecallfocallossforimageclassificationandsegmentation
AT lijingma drfldynamicrecallfocallossforimageclassificationandsegmentation
AT chaoliwang drfldynamicrecallfocallossforimageclassificationandsegmentation