Lateral Feature Enhancement Network for Page Object Detection

In this article, a lateral feature enhancement (LFE) backbone network is proposed to enrich feature representation effectively for page object detection (POD) across various scales. Our LFE backbone network has three feature enhancement modules. First, feature enhancement of large page object is a bottom-up feature pyramid, enhancing features of large page objects, which convey more important information to readers. Second, the LFE includes a top-down feature pyramid propagating representative semantical features to lower layers and a lateral connection for feature enhancement in each layer. Third, lateral skip connection is designed to retain the original feature details. The stacking strategies of bottom-up, top-down, and lateral connections are beneficial to overall object detection. Visualization of feature indicates that the proposed LFE backbone network enhances global semantic information as well as detailed features of small page objects. Comparative experiments on the two state-of-the-art datasets show that it achieves excellent results with 0.950 mean of AP (mAP) on PubLayNet and 0.892 mAP on POD with more strict metric intersection over union (IoU)
$=0.8$
, respectively. Compared with both computer vision (CV)-based unimodal detectors and multimodal detectors, the proposed LFE network performs excellently. Visual effect experiments compare the performances of CV-based detectors. The results show that our detector outperforms others with strict metric, especially in the detection of small page objects.