Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images

In object detection on remote sensing images or aerial images, high-resolution images and low relative area ratio of objects need to be solved. Usually, a high-resolution image should be split and detected separately. Then, the prediction results would be merged as a result of the complete image. Du...

Full description

Saved in:
Bibliographic Details
Main Authors: Lei Ge, Lei Dou
Format: Article
Language:English
Published: IEEE 2024-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10701299/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850283085192495104
author Lei Ge
Lei Dou
author_facet Lei Ge
Lei Dou
author_sort Lei Ge
collection DOAJ
description In object detection on remote sensing images or aerial images, high-resolution images and low relative area ratio of objects need to be solved. Usually, a high-resolution image should be split and detected separately. Then, the prediction results would be merged as a result of the complete image. Due to the overlap between adjacent image slices, non-maximum suppression (NMS) is used to suppress redundant prediction boxes during the merging process. However, redundant boxes from splitting are different from those generated by detection on a single image. Some special cases cannot be effectively solved by NMS. We have studied these cases and summarized judgment conditions to identify them. Based on NMS, we propose a new method named RODM-NMS for rotated object detection during merging slices of high-resolution images. Cases challenging for NMS are classified into two categories and addressed with specific handling methods. Compared to the traditional NMS, RODM-NMS could achieve a 3.9 improvement in AP50 during merging detection results on <inline-formula> <tex-math notation="LaTeX">$640\times 640$ </tex-math></inline-formula> slices of DOTA by FCOSR. It has low requirements for the overlap gap of slices and shows a significant improvement in performance for weak models. Hence, it is more suitable for computing-constrained mobile platforms such as drones to detect objects on high-resolution images.
format Article
id doaj-art-15ccdd285ffe4c94afa5292e55b19139
institution OA Journals
issn 2169-3536
language English
publishDate 2024-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-15ccdd285ffe4c94afa5292e55b191392025-08-20T01:47:50ZengIEEEIEEE Access2169-35362024-01-011214999915000710.1109/ACCESS.2024.347081510701299Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution ImagesLei Ge0https://orcid.org/0000-0002-5697-0087Lei Dou1https://orcid.org/0000-0002-8889-6862National Key Laboratory of Transient Physics, Nanjing University of Science and Technology, Nanjing, ChinaNational Key Laboratory of Transient Physics, Nanjing University of Science and Technology, Nanjing, ChinaIn object detection on remote sensing images or aerial images, high-resolution images and low relative area ratio of objects need to be solved. Usually, a high-resolution image should be split and detected separately. Then, the prediction results would be merged as a result of the complete image. Due to the overlap between adjacent image slices, non-maximum suppression (NMS) is used to suppress redundant prediction boxes during the merging process. However, redundant boxes from splitting are different from those generated by detection on a single image. Some special cases cannot be effectively solved by NMS. We have studied these cases and summarized judgment conditions to identify them. Based on NMS, we propose a new method named RODM-NMS for rotated object detection during merging slices of high-resolution images. Cases challenging for NMS are classified into two categories and addressed with specific handling methods. Compared to the traditional NMS, RODM-NMS could achieve a 3.9 improvement in AP50 during merging detection results on <inline-formula> <tex-math notation="LaTeX">$640\times 640$ </tex-math></inline-formula> slices of DOTA by FCOSR. It has low requirements for the overlap gap of slices and shows a significant improvement in performance for weak models. Hence, it is more suitable for computing-constrained mobile platforms such as drones to detect objects on high-resolution images.https://ieeexplore.ieee.org/document/10701299/Non-maximum suppressionrotated object detectionremote sensingsmall object detection
spellingShingle Lei Ge
Lei Dou
Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
IEEE Access
Non-maximum suppression
rotated object detection
remote sensing
small object detection
title Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
title_full Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
title_fullStr Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
title_full_unstemmed Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
title_short Non-Maximum Suppression for Rotated Object Detection During Merging Slices of High-Resolution Images
title_sort non maximum suppression for rotated object detection during merging slices of high resolution images
topic Non-maximum suppression
rotated object detection
remote sensing
small object detection
url https://ieeexplore.ieee.org/document/10701299/
work_keys_str_mv AT leige nonmaximumsuppressionforrotatedobjectdetectionduringmergingslicesofhighresolutionimages
AT leidou nonmaximumsuppressionforrotatedobjectdetectionduringmergingslicesofhighresolutionimages