Improving the Performance of Object Detection by Preserving Balanced Class Distribution
Object detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and va...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-10-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/11/21/4460 |
| _version_ | 1797631636533149696 |
|---|---|
| author | Heewon Lee Sangtae Ahn |
| author_facet | Heewon Lee Sangtae Ahn |
| author_sort | Heewon Lee |
| collection | DOAJ |
| description | Object detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and validation typically originate from public datasets that are well-balanced in terms of the number of objects ascribed to each class in an image. However, in real-world scenarios, handling datasets with much greater class imbalance, i.e., very different numbers of objects for each class, is much more common, and this imbalance may reduce the performance of object detection when predicting unseen test images. In our study, thus, we propose a method that evenly distributes the classes in an image for training and validation, solving the class imbalance problem in object detection. Our proposed method aims to maintain a uniform class distribution through multi-label stratification. We tested our proposed method not only on public datasets that typically exhibit balanced class distribution but also on private datasets that may have imbalanced class distribution. We found that our proposed method was more effective on datasets containing severe imbalance and less data. Our findings indicate that the proposed method can be effectively used on datasets with substantially imbalanced class distribution. |
| first_indexed | 2024-03-11T11:25:15Z |
| format | Article |
| id | doaj.art-548c8cb56bb94492a7b744327db0b6fe |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-11T11:25:15Z |
| publishDate | 2023-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-548c8cb56bb94492a7b744327db0b6fe2023-11-10T15:07:57ZengMDPI AGMathematics2227-73902023-10-011121446010.3390/math11214460Improving the Performance of Object Detection by Preserving Balanced Class DistributionHeewon Lee0Sangtae Ahn1School of Electronics Engineering, Kyungpook National University, Daegu 41566, Republic of KoreaSchool of Electronics Engineering, Kyungpook National University, Daegu 41566, Republic of KoreaObject detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and validation typically originate from public datasets that are well-balanced in terms of the number of objects ascribed to each class in an image. However, in real-world scenarios, handling datasets with much greater class imbalance, i.e., very different numbers of objects for each class, is much more common, and this imbalance may reduce the performance of object detection when predicting unseen test images. In our study, thus, we propose a method that evenly distributes the classes in an image for training and validation, solving the class imbalance problem in object detection. Our proposed method aims to maintain a uniform class distribution through multi-label stratification. We tested our proposed method not only on public datasets that typically exhibit balanced class distribution but also on private datasets that may have imbalanced class distribution. We found that our proposed method was more effective on datasets containing severe imbalance and less data. Our findings indicate that the proposed method can be effectively used on datasets with substantially imbalanced class distribution.https://www.mdpi.com/2227-7390/11/21/4460computer visionobject detectionimbalanced class distributionmulti-label stratification |
| spellingShingle | Heewon Lee Sangtae Ahn Improving the Performance of Object Detection by Preserving Balanced Class Distribution Mathematics computer vision object detection imbalanced class distribution multi-label stratification |
| title | Improving the Performance of Object Detection by Preserving Balanced Class Distribution |
| title_full | Improving the Performance of Object Detection by Preserving Balanced Class Distribution |
| title_fullStr | Improving the Performance of Object Detection by Preserving Balanced Class Distribution |
| title_full_unstemmed | Improving the Performance of Object Detection by Preserving Balanced Class Distribution |
| title_short | Improving the Performance of Object Detection by Preserving Balanced Class Distribution |
| title_sort | improving the performance of object detection by preserving balanced class distribution |
| topic | computer vision object detection imbalanced class distribution multi-label stratification |
| url | https://www.mdpi.com/2227-7390/11/21/4460 |
| work_keys_str_mv | AT heewonlee improvingtheperformanceofobjectdetectionbypreservingbalancedclassdistribution AT sangtaeahn improvingtheperformanceofobjectdetectionbypreservingbalancedclassdistribution |