Open Access

Fig. 4

image

Download original image

Specific structure of the YOLO-DG network. The network structure can be divided into three parts: the Input module describes the image data, image dimensions, and preprocessing (Data augmentation); the Backbone and Head parts are the specific structure of YOLO-DG; the Result module is a visual display of the YOLO-DG result. The Structure of SPD-Conv module and the Structure of coordinate attention (CA) module below the figure respectively show the detailed structure of the modules we added. SPD-Conv consists of a space-to-depth (SPD) layer and a non-strided convolution layer, and CA consists of two parts: coordinate information embedding and coordinate attention generation.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.