LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection
This paper proposes a learnable line encoding technique for bounding boxes commonly used in the object detection task. A bounding box is simply encoded using two main points: the top-left corner and the bottom-right corner of the bounding box; then, a lightweight convolutional neural network (CNN) i...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-05-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/10/3699 |
| _version_ | 1797495737440796672 |
|---|---|
| author | Hatem Ibrahem Ahmed Salem Hyun-Soo Kang |
| author_facet | Hatem Ibrahem Ahmed Salem Hyun-Soo Kang |
| author_sort | Hatem Ibrahem |
| collection | DOAJ |
| description | This paper proposes a learnable line encoding technique for bounding boxes commonly used in the object detection task. A bounding box is simply encoded using two main points: the top-left corner and the bottom-right corner of the bounding box; then, a lightweight convolutional neural network (CNN) is employed to learn the lines and propose high-resolution line masks for each category of classes using a pixel-shuffle operation. Post-processing is applied to the predicted line masks to filtrate them and estimate clear lines based on a progressive probabilistic Hough transform. The proposed method was trained and evaluated on two common object detection benchmarks: Pascal VOC2007 and MS-COCO2017. The proposed model attains high mean average precision (mAP) values (78.8% for VOC2007 and 48.1% for COCO2017) while processing each frame in a few milliseconds (37 ms for PASCAL VOC and 47 ms for COCO). The strength of the proposed method lies in its simplicity and ease of implementation unlike the recent state-of-the-art methods in object detection, which include complex processing pipelines. |
| first_indexed | 2024-03-10T01:53:54Z |
| format | Article |
| id | doaj.art-4da92afa212d4907928d724a46f2ac19 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T01:53:54Z |
| publishDate | 2022-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-4da92afa212d4907928d724a46f2ac192023-11-23T12:59:39ZengMDPI AGSensors1424-82202022-05-012210369910.3390/s22103699LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object DetectionHatem Ibrahem0Ahmed Salem1Hyun-Soo Kang2Department of Information and Communication Engineering, School of Electrical and Computer Engineering, Chungbuk National University, Cheongju-si 28644, KoreaDepartment of Information and Communication Engineering, School of Electrical and Computer Engineering, Chungbuk National University, Cheongju-si 28644, KoreaDepartment of Information and Communication Engineering, School of Electrical and Computer Engineering, Chungbuk National University, Cheongju-si 28644, KoreaThis paper proposes a learnable line encoding technique for bounding boxes commonly used in the object detection task. A bounding box is simply encoded using two main points: the top-left corner and the bottom-right corner of the bounding box; then, a lightweight convolutional neural network (CNN) is employed to learn the lines and propose high-resolution line masks for each category of classes using a pixel-shuffle operation. Post-processing is applied to the predicted line masks to filtrate them and estimate clear lines based on a progressive probabilistic Hough transform. The proposed method was trained and evaluated on two common object detection benchmarks: Pascal VOC2007 and MS-COCO2017. The proposed model attains high mean average precision (mAP) values (78.8% for VOC2007 and 48.1% for COCO2017) while processing each frame in a few milliseconds (37 ms for PASCAL VOC and 47 ms for COCO). The strength of the proposed method lies in its simplicity and ease of implementation unlike the recent state-of-the-art methods in object detection, which include complex processing pipelines.https://www.mdpi.com/1424-8220/22/10/3699object detectionconvolutional neural networksline detectionreal-time processing |
| spellingShingle | Hatem Ibrahem Ahmed Salem Hyun-Soo Kang LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection Sensors object detection convolutional neural networks line detection real-time processing |
| title | LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection |
| title_full | LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection |
| title_fullStr | LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection |
| title_full_unstemmed | LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection |
| title_short | LEOD-Net: Learning Line-Encoded Bounding Boxes for Real-Time Object Detection |
| title_sort | leod net learning line encoded bounding boxes for real time object detection |
| topic | object detection convolutional neural networks line detection real-time processing |
| url | https://www.mdpi.com/1424-8220/22/10/3699 |
| work_keys_str_mv | AT hatemibrahem leodnetlearninglineencodedboundingboxesforrealtimeobjectdetection AT ahmedsalem leodnetlearninglineencodedboundingboxesforrealtimeobjectdetection AT hyunsookang leodnetlearninglineencodedboundingboxesforrealtimeobjectdetection |