Multi-Scale Infrared Pedestrian Detection Based on Deep Attention Mechanism

Bin, 赵斌 Zhao; Chunping, 王春平 Wang; Qiang, 付强 Fu; Yichao, 陈一超 Chen

doi:10.3788/aos202040.0504001

Cited by 5 publications

(2 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent studies have shown that convolutional neural networks can automatically mine the hidden statistical laws and essential features from infrared images, which makes the detector's detection effect of pedestrian objects far better than traditional detection methods 1 . The general development trend of the two-stage network is to design deeper and more complex networks to fully extract the characteristic information of pedestrians in infrared images, trying to obtain higher detection accuracy.…”

Section: Introductionmentioning

confidence: 99%

“…Nevertheless, this operation destroys the original structure of the model and results in a decrease in detection accuracy. He et al 5 and Zhang et al 6 used additional pruning and quantization strategies to improve the inference speed, but this method increased the algorithm complexity. Therefore, how to conduct real-time and effective detection of road infrared pedestrian objects is still one of the most challenging tasks faced by ADAS.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Real-time detection method for infrared pedestrian

Chen,

Liu

2023

Fourth International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2023)

View full text Add to dashboard Cite

We propose a real-time infrared pedestrian detection method, which achieves higher performance than the state-of-theart YOLO series in infrared pedestrian detection. Our method is based on the YOLOv6-n with some new technologies. In particular, Mixed Downsampling Module (MDM) and Cross-stage Connection Structure (CSCS) are designed to reduce the inference latency of the network. In addition, Simplified Spatial Attention Module (SimSAM) is proposed to improve detection performance with negligible overheads. We validate our method through extensive experiments on FLIR detection datasets. Experimental results show that our method can achieve 84.86% AP with inference latency of 0.54 ms on an NVIDIA 1080Ti GPU.

show abstract

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%