Improving YOLOv4-Tiny’s Construction Machinery and Material Identification Method by Incorporating Attention Mechanism

Abstract

To facilitate the development of intelligent unmanned loaders and improve the recognition accuracy of loaders in complex scenes, we propose a construction machinery and material target detection algorithm incorporating an attention mechanism (AM) to improve YOLOv4-Tiny. First, to ensure the robustness of the proposed algorithm, we adopt style migration and sliding window segmentation to increase the underlying dataset’s diversity. Second, to address the problem that YOLOv4-Tiny’s (the base network) framework only adopts a layer-by-layer connection form, which demonstrates an insufficient feature extraction ability, we adopt a multilayer cascaded residual module to deeply connect low- and high-level information. Finally, to filter redundant feature information and make the proposed algorithm focus more on important feature information, a channel AM is added to the base network to perform a secondary screening of feature information in the region of interest, which effectively improves the detection accuracy. In addition, to achieve small-scale object detection, a multiscale feature pyramid network structure is employed in the prediction module of the proposed algorithm to output two prediction networks with different scale sizes. The experimental results show that, compared with the traditional network structure, the proposed algorithm fully incorporates the advantages of residual networks and AM, which effectively improves its feature extraction ability and recognition accuracy of targets at different scales. The final proposed algorithm exhibits the features of high recognition accuracy and fast recognition speed, with mean average precision and detection speed reaching 96.82% and 134.4 fps, respectively.

Improved YOLOv4-tiny Target Detection Method Based on Adaptive Self-Order Piecewise Enhancement and Multiscale Feature Optimization

“…where r represents the pixel value of the image and MSE represents the mean squared error. Among the target detection accuracy evaluation indicators, the accuracy, recall, F1 score, and average accuracy (mAP) evaluation metrics are as follows [29]:…”

Vehicle door frame positioning method for binocular vision robots based on improved YOLOv4

“…The large size and computational complexity of YOLOv4 and other network models make it difficult to run with low performance devices and are not frequently used in the field of automotive door frame inspection. Although the YOLOv4-Tiny model is small in size and computation, the detection accuracy of the model is poor and cannot be adapted to the process of automotive door frame recognition in complicated environments [11]. In this paper, we propose a lightweight and efficient feature fusion target detection network (LEFFDet) for automotive door frame recognition in complex environments.…”

“…Guo et al [12] proposed a fast detection algorithm for metal parts surface defects based on the YOLOv4-mobilenet network by designing a feature extraction network with faster detection speed based on MobileNetv3 with depth separable convolution, but the detection accuracy is low. Yao et al [11] enhanced YOLOv4-Tiny by using a multilayer cascaded residual module and incorporating an attention mechanism for engineering machinery materials to refine the detection accuracy. Although these studies optimized the network model, they did not consider the effect of ambient light changes on the detection effect in the complex working environment, especially in the complex working environment.…”

Automatic recognition of parasitic products in stool examination using object detection approach

“…Due to the lower accuracy of YOLOv3, larger weight of YOLOv4, and complex network structure of YOLOv5, a more simplified version of YOLOv4 (YOLOv4-Tiny) was designed to maximize detection speed and improve computational efficiency ( Wang, Bochkovskiy & Liao, 2021 ). In particular, it has been applied to the pine wilt disease detection ( Li et al, 2021 ), trash detection ( Kulshreshtha et al, 2021 ), multi-object tracking ( Wu et al, 2021 ), electronic component detection ( Guo et al, 2021 ), construction machinery and material identification ( Yao et al, 2022 ), and fruit flies gender classification ( Genaev et al, 2022 ). Based on the literatures, our research also focused on the fourth tiny model of YOLO, namely the YOLOv4-Tiny because it produces faster detection results and uses less memory with the support of low-end GPU devices.…”