Detection of Small Ship Objects Using Anchor Boxes Cluster and Feature Pyramid Network Model for SAR Imagery
FPN: suitable for the detection of multiple small ship objects in SAR images.
论文目的
This study aims to resolve the problems associated with small-object and multi-object ship detection in complex scenarios by proposing a detection method based on an optimized FPN model.
论文算法
1.embedded in a traditional region proposal network (RPN) and mapped into a new feature space for object identification.
2.the k-means clustering algorithm based on the shape similar distance (SSD) measure is used to optimize the FPN
论文中提出由于目标舰船是多尺度的,所以使用anchor-based算法生成FPN map.由于Anchor的数量和尺度是一组超参数,并且在实际中目标非常容易变化,因此anchor生成机制会减缓收敛,因此选择初始化anchor boxes。即使用ssd-kmeans进行聚类。这里的ssd指的是shape similar distance
论文结果
shows an accuracyof98.62%.
创新点
ssd-kmeans
缺点
首先是没有实时性和参数对比,其次,针对复杂场景,误检率没有提高。