Dense-RefineDet for Traffic Sign Detection and Classification
Detecting and classifying real-life small traffic signs from large input images is difficult due to their occupying fewer pixels relative to larger targets. To address this challenge, we proposed a deep-learning-based model (Dense-RefineDet) that applies a single-shot, object-detection framework (Re...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-11-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/20/22/6570 |
| _version_ | 1797547661272809472 |
|---|---|
| author | Chang Sun Yibo Ai Sheng Wang Weidong Zhang |
| author_facet | Chang Sun Yibo Ai Sheng Wang Weidong Zhang |
| author_sort | Chang Sun |
| collection | DOAJ |
| description | Detecting and classifying real-life small traffic signs from large input images is difficult due to their occupying fewer pixels relative to larger targets. To address this challenge, we proposed a deep-learning-based model (Dense-RefineDet) that applies a single-shot, object-detection framework (RefineDet) to maintain a suitable accuracy–speed trade-off. We constructed a dense connection-related transfer-connection block to combine high-level feature layers with low-level feature layers to optimize the use of the higher layers to obtain additional contextual information. Additionally, we presented an anchor-design method to provide suitable anchors for detecting small traffic signs. Experiments using the Tsinghua-Tencent 100K dataset demonstrated that Dense-RefineDet achieved competitive accuracy at high-speed detection (0.13 s/frame) of small-, medium-, and large-scale traffic signs (recall: 84.3%, 95.2%, and 92.6%; precision: 83.9%, 95.6%, and 94.0%). Moreover, experiments using the Caltech pedestrian dataset indicated that the miss rate of Dense-RefineDet was 54.03% (pedestrian height > 20 pixels), which outperformed other state-of-the-art methods. |
| first_indexed | 2024-03-10T14:47:17Z |
| format | Article |
| id | doaj.art-ead33f1f4c4442428299223f5f2a08fa |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T14:47:17Z |
| publishDate | 2020-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-ead33f1f4c4442428299223f5f2a08fa2023-11-20T21:17:29ZengMDPI AGSensors1424-82202020-11-012022657010.3390/s20226570Dense-RefineDet for Traffic Sign Detection and ClassificationChang Sun0Yibo Ai1Sheng Wang2Weidong Zhang3National Center for Materials Service Safety, University of Science and Technology Beijing, Beijing 100083, ChinaNational Center for Materials Service Safety, University of Science and Technology Beijing, Beijing 100083, ChinaAI Lab, UCAR, 118 East Zhongguancun Road, Haidian District, Beijing 100098, ChinaNational Center for Materials Service Safety, University of Science and Technology Beijing, Beijing 100083, ChinaDetecting and classifying real-life small traffic signs from large input images is difficult due to their occupying fewer pixels relative to larger targets. To address this challenge, we proposed a deep-learning-based model (Dense-RefineDet) that applies a single-shot, object-detection framework (RefineDet) to maintain a suitable accuracy–speed trade-off. We constructed a dense connection-related transfer-connection block to combine high-level feature layers with low-level feature layers to optimize the use of the higher layers to obtain additional contextual information. Additionally, we presented an anchor-design method to provide suitable anchors for detecting small traffic signs. Experiments using the Tsinghua-Tencent 100K dataset demonstrated that Dense-RefineDet achieved competitive accuracy at high-speed detection (0.13 s/frame) of small-, medium-, and large-scale traffic signs (recall: 84.3%, 95.2%, and 92.6%; precision: 83.9%, 95.6%, and 94.0%). Moreover, experiments using the Caltech pedestrian dataset indicated that the miss rate of Dense-RefineDet was 54.03% (pedestrian height > 20 pixels), which outperformed other state-of-the-art methods.https://www.mdpi.com/1424-8220/20/22/6570deep learningneural networkobject detectiontraffic sign recognitiondense connectionanchor design |
| spellingShingle | Chang Sun Yibo Ai Sheng Wang Weidong Zhang Dense-RefineDet for Traffic Sign Detection and Classification Sensors deep learning neural network object detection traffic sign recognition dense connection anchor design |
| title | Dense-RefineDet for Traffic Sign Detection and Classification |
| title_full | Dense-RefineDet for Traffic Sign Detection and Classification |
| title_fullStr | Dense-RefineDet for Traffic Sign Detection and Classification |
| title_full_unstemmed | Dense-RefineDet for Traffic Sign Detection and Classification |
| title_short | Dense-RefineDet for Traffic Sign Detection and Classification |
| title_sort | dense refinedet for traffic sign detection and classification |
| topic | deep learning neural network object detection traffic sign recognition dense connection anchor design |
| url | https://www.mdpi.com/1424-8220/20/22/6570 |
| work_keys_str_mv | AT changsun denserefinedetfortrafficsigndetectionandclassification AT yiboai denserefinedetfortrafficsigndetectionandclassification AT shengwang denserefinedetfortrafficsigndetectionandclassification AT weidongzhang denserefinedetfortrafficsigndetectionandclassification |