An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window
Drone object detection faces numerous challenges such as dense clusters with overlapping, scale diversity, and long-tail distributions. Utilizing tiling inference through uniform sliding window is an effective way of enlarging tiny objects and meanwhile efficient for real-world applications. However...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/17/4122 |
| _version_ | 1797581902539915264 |
|---|---|
| author | Xin Yang Yong Song Ya Zhou Yizhao Liao Jinqi Yang Jinxiang Huang Yiqian Huang Yashuo Bai |
| author_facet | Xin Yang Yong Song Ya Zhou Yizhao Liao Jinqi Yang Jinxiang Huang Yiqian Huang Yashuo Bai |
| author_sort | Xin Yang |
| collection | DOAJ |
| description | Drone object detection faces numerous challenges such as dense clusters with overlapping, scale diversity, and long-tail distributions. Utilizing tiling inference through uniform sliding window is an effective way of enlarging tiny objects and meanwhile efficient for real-world applications. However, merely partitioning input images may result in heavy truncation and an unexpected performance drop in large objects. Therefore, in this work, we strive to develop an improved tiling detection framework with both competitive performance and high efficiency. First, we formulate the tiling inference and training pipeline with a mixed data strategy. To avoid truncation and handle objects at all scales, we simultaneously perform global detection on the original image and local detection on corresponding sub-patches, employing appropriate patch settings. Correspondingly, the training data includes both original images and the patches generated by random online anchor-cropping, which can ensure the effectiveness of patches and enrich the image scenarios. Furthermore, a scale filtering mechanism is applied to assign objects at diverse scales to global and local detection tasks to keep the scale invariance of a detector and obtain optimal fused predictions. As most of the additional operations are performed in parallel, the tiling inference remains highly efficient. Additionally, we devise two augmentations customized for tiling detection to effectively increase valid annotations, which can generate more challenging drone scenarios and simulate the practical cluster with overlapping, especially for rare categories. Comprehensive experiments on both public drone benchmarks and our customized real-world images demonstrate that, in comparison to other drone detection frameworks, the proposed tiling framework can significantly improve the performance of general detectors in drone scenarios with lower additional computational costs. |
| first_indexed | 2024-03-10T23:14:14Z |
| format | Article |
| id | doaj.art-46957ab64c5947b5ae3d3613291642e9 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T23:14:14Z |
| publishDate | 2023-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-46957ab64c5947b5ae3d3613291642e92023-11-19T08:44:43ZengMDPI AGRemote Sensing2072-42922023-08-011517412210.3390/rs15174122An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing WindowXin Yang0Yong Song1Ya Zhou2Yizhao Liao3Jinqi Yang4Jinxiang Huang5Yiqian Huang6Yashuo Bai7School of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaSchool of Optics and Photonics, Beijing Institute of Technology, Beijing 100081, ChinaDrone object detection faces numerous challenges such as dense clusters with overlapping, scale diversity, and long-tail distributions. Utilizing tiling inference through uniform sliding window is an effective way of enlarging tiny objects and meanwhile efficient for real-world applications. However, merely partitioning input images may result in heavy truncation and an unexpected performance drop in large objects. Therefore, in this work, we strive to develop an improved tiling detection framework with both competitive performance and high efficiency. First, we formulate the tiling inference and training pipeline with a mixed data strategy. To avoid truncation and handle objects at all scales, we simultaneously perform global detection on the original image and local detection on corresponding sub-patches, employing appropriate patch settings. Correspondingly, the training data includes both original images and the patches generated by random online anchor-cropping, which can ensure the effectiveness of patches and enrich the image scenarios. Furthermore, a scale filtering mechanism is applied to assign objects at diverse scales to global and local detection tasks to keep the scale invariance of a detector and obtain optimal fused predictions. As most of the additional operations are performed in parallel, the tiling inference remains highly efficient. Additionally, we devise two augmentations customized for tiling detection to effectively increase valid annotations, which can generate more challenging drone scenarios and simulate the practical cluster with overlapping, especially for rare categories. Comprehensive experiments on both public drone benchmarks and our customized real-world images demonstrate that, in comparison to other drone detection frameworks, the proposed tiling framework can significantly improve the performance of general detectors in drone scenarios with lower additional computational costs.https://www.mdpi.com/2072-4292/15/17/4122aerial object detectionsliding windowaugmentationunmanned aerial vehicles |
| spellingShingle | Xin Yang Yong Song Ya Zhou Yizhao Liao Jinqi Yang Jinxiang Huang Yiqian Huang Yashuo Bai An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window Remote Sensing aerial object detection sliding window augmentation unmanned aerial vehicles |
| title | An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window |
| title_full | An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window |
| title_fullStr | An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window |
| title_full_unstemmed | An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window |
| title_short | An Efficient Detection Framework for Aerial Imagery Based on Uniform Slicing Window |
| title_sort | efficient detection framework for aerial imagery based on uniform slicing window |
| topic | aerial object detection sliding window augmentation unmanned aerial vehicles |
| url | https://www.mdpi.com/2072-4292/15/17/4122 |
| work_keys_str_mv | AT xinyang anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yongsong anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yazhou anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yizhaoliao anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT jinqiyang anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT jinxianghuang anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yiqianhuang anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yashuobai anefficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT xinyang efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yongsong efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yazhou efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yizhaoliao efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT jinqiyang efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT jinxianghuang efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yiqianhuang efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow AT yashuobai efficientdetectionframeworkforaerialimagerybasedonuniformslicingwindow |