A Shallow Pooled Weighted Feature Enhancement Network for Small-Sized Pine Wilt Diseased Tree Detection

Pine wild disease poses a serious threat to the ecological environment of national forests. Combining object detection algorithm with Unmanned Aerial Vehicles (UAV) detect pine diseased trees (PWDT) is significant step in preventing spread disease. To address issue shallow feature layers lacking ability fully extract features from small-sized existing algorithms, as well problem small number single image, Shallow Pooled Weighted Feature Enhancement Network (SPW-FEN) based on Small Target Expansion (STE) has been proposed for detecting PWDT. First, Channel Attention (PWCA) module presented and introduced into layer rich target information enhance network’s expressive regarding characteristics two-layer maps. Additionally, an STE data enhancement method targets, which effectively increases sample size image. The experimental results PWDT dataset indicate that achieved average precision recall 79.1% 86.9%, respectively. This 3.6 3.8 percentage points higher, respectively, than recognition state-of-the-art Faster-RCNN, 6.4 5.5 higher those newly YOLOv6 method.