Research on pedestrian detection based on multi-level fine-grained YOLOX algorithm

Wang, Hong; Xie, Yuan; Tian, Shasha; Li, Zheng; Dong, Xiaojie; Zhu, Yan

doi:10.1108/ijicc-05-2022-0161

Cited by 5 publications

(4 citation statements)

References 52 publications

(61 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among the current mainstream one-stage detection methods, the YOLO series is divided into multiple versions, including YOLOv3 [7], YOLOv4 [8], YOLOv5 [9], YOLOX [10], and YOLO-tiny [11]. Similarly, multiple improved versions of SSD [12] have been launched.…”

Section: One-stage Detection Methodsmentioning

confidence: 99%

CaSnLi-YOLO: construction site multi-target detection method based on improved YOLOv5s

Hao,

Zhang,

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

To address the problem of the frequent occurrence of major casualties during construction, a lightweight multi-target detection model based on YOLOv5s, named CaSnLi-YOLO, was proposed for the detection of multiple targets consisting of construction workers and various mechanical equipment at construction sites. In this study, the introduction of the Coordinate Attention (CA) mechanism along with DWConv and C3Ghost based on GhostNet, enhanced the expression and detection accuracy of the model, effectively reducing the number of parameters. Considering the environmental characteristics of construction sites, a detection box filtering strategy based on Soft Non-Maximum Suppression (Soft-NMS) was employed, to effectively alleviate the issue of missed detections of occluded targets. Experimental results demonstrate the significant superiority of the proposed CaSnLi-YOLO over current mainstream detection models, such as faster region-based convolutional neural network (Faster R-CNN) and single-shot detector (SSD). The proposed CaSnLi-YOLO has a parameter number of 5.96×106, which is 15.2% less than that of the original YOLOv5s model, further improving precision, recall rate, mAP@0.5, and mAP@0.5:0.95 by 0.6%, 0.6 %, 0.2%, and 2.3%, respectively. The improved YOLOv5s model proposed in this study achieved significant improvements in multi-target detection at construction sites. The model demonstrated effective enhancements while significantly reducing parameter count and is expected to be deployed in small-edge devices for real-time security monitoring at construction sites.

show abstract

Section: One-stage Detection Methodsmentioning

confidence: 99%

CaSnLi-YOLO: construction site multi-target detection method based on improved YOLOv5s

Hao,

Zhang,

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

show abstract

“…, 2022a) and is publicly available on the DarkNet repository (Alexey, 2020). The YOLO models have shown fascinating results in many downstream tasks (Wang et al. , 2022b; Yang et al.…”

Section: Methodsmentioning

confidence: 99%

“…The Tiny YOLOv7 is a recent addition to the series of Tiny YOLO object detectors proposed in (Wang et al, 2022a) and is publicly available on the DarkNet repository (Alexey, 2020). The YOLO models have shown fascinating results in many downstream tasks (Wang et al, 2022b;Yang et al, 2023). The Tiny YOLOv7 is made up of 72 convolutional layers and three detection heads, as shown in Figure 2.…”

Section: Proposed Eye-yolomentioning

confidence: 99%

EYE-YOLO: a multi-spatial pyramid pooling and Focal-EIOU loss inspired tiny YOLOv7 for fundus eye disease detection

Kumar,

Dhanalakshmi

2024

IJICC

View full text Add to dashboard Cite

PurposeThe purpose of this work is to present an approach for autonomous detection of eye disease in fundus images. Furthermore, this work presents an improved variant of the Tiny YOLOv7 model developed specifically for eye disease detection. The model proposed in this work is a highly useful tool for the development of applications for autonomous detection of eye diseases in fundus images that can help and assist ophthalmologists.Design/methodology/approachThe approach adopted to carry out this work is twofold. Firstly, a richly annotated dataset consisting of eye disease classes, namely, cataract, glaucoma, retinal disease and normal eye, was created. Secondly, an improved variant of the Tiny YOLOv7 model was developed and proposed as EYE-YOLO. The proposed EYE-YOLO model has been developed by integrating multi-spatial pyramid pooling in the feature extraction network and Focal-EIOU loss in the detection network of the Tiny YOLOv7 model. Moreover, at run time, the mosaic augmentation strategy has been utilized with the proposed model to achieve benchmark results. Further, evaluations have been carried out for performance metrics, namely, precision, recall, F1 Score, average precision (AP) and mean average precision (mAP).FindingsThe proposed EYE-YOLO achieved 28% higher precision, 18% higher recall, 24% higher F1 Score and 30.81% higher mAP than the Tiny YOLOv7 model. Moreover, in terms of AP for each class of the employed dataset, it achieved 9.74% higher AP for cataract, 27.73% higher AP for glaucoma, 72.50% higher AP for retina disease and 13.26% higher AP for normal eye. In comparison to the state-of-the-art Tiny YOLOv5, Tiny YOLOv6 and Tiny YOLOv8 models, the proposed EYE-YOLO achieved 6–23.32% higher mAP.Originality/valueThis work addresses the problem of eye disease recognition as a bounding box regression and detection problem. Whereas, the work in the related research is largely based on eye disease classification. The other highlight of this work is to propose a richly annotated dataset for different eye diseases useful for training deep learning-based object detectors. The major highlight of this work lies in the proposal of an improved variant of the Tiny YOLOv7 model focusing on eye disease detection. The proposed modifications in the Tiny YOLOv7 aided the proposed model in achieving better results as compared to the state-of-the-art Tiny YOLOv8 and YOLOv8 Nano.

show abstract

“…Feature Pyramid Network (FPN) [34][35][36] in the feature aggregation stage, pyramid feature maps of different scales are usually obtained along the bottom-up path. This way, when the shallow features are transferred to the top layer, they must go through multiple network layers, resulting in low-level features.…”

Section: Criss-cross Fusion Module (Ccfm)mentioning

confidence: 99%

LAGSwin: Local attention guided Swin-transformer for thermal infrared sports object detection

Meng,

Si,

Mao

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

Compared with visible light images, thermal infrared images have poor resolution, low contrast, signal-to-noise ratio, blurred visual effects, and less information. Thermal infrared sports target detection methods relying on traditional convolutional networks capture the rich semantics in high-level features but blur the spatial details. The differences in physical information content and spatial distribution of high and low features are ignored, resulting in a mismatch between the region of interest and the target. To address these issues, we propose a local attention-guided Swin-transformer thermal infrared sports object detection method (LAGSwin) to encode sports objects’ spatial transformation and orientation information. On the one hand, Swin-transformer guided by local attention is adopted to enrich the semantic knowledge of low-level features by embedding local focus from high-level features and generating high-quality anchors while increasing the embedding of contextual information. On the other hand, an active rotation filter is employed to encode orientation information, resulting in orientation-sensitive and invariant features to reduce the inconsistency between classification and localization regression. A bidirectional criss-cross fusion strategy is adopted in the feature fusion stage to enable better interaction and embedding features of different resolutions. At last, the evaluation and verification of multiple open-source sports target datasets prove that the proposed LAGSwin detection framework has good robustness and generalization ability.

show abstract

Research on pedestrian detection based on multi-level fine-grained YOLOX algorithm

Cited by 5 publications

References 52 publications

CaSnLi-YOLO: construction site multi-target detection method based on improved YOLOv5s

CaSnLi-YOLO: construction site multi-target detection method based on improved YOLOv5s

EYE-YOLO: a multi-spatial pyramid pooling and Focal-EIOU loss inspired tiny YOLOv7 for fundus eye disease detection

LAGSwin: Local attention guided Swin-transformer for thermal infrared sports object detection

Contact Info

Product

Resources

About