作者:Jingru Yi, Pengxiang Wu, Bo Liu, Qiaoying Huang, Hui Qu, Dimitris Metaxas
发布时间:2020
发布期刊:Arixv
论文全称:Oriented Object Detection in Aerial Images with Box Boundary-Aware Vectors
论文地址:https://paperswithcode.com/paper/oriented-object-detection-in-aerial-images
代码:https://github.com/yijingru/BBAVectors-Oriented-Object-Detection
Oriented object detection in aerial images is a challenging task as the objects in aerial images are displayed in arbitrary directions and are usually densely packed. Current oriented object detection methods mainly rely on two stage anchor-based detectors. However, the anchor-based detectors typically suffer from a severe imbalance issue between the positive and negative anchor boxes. To address this issue, in this work we extend the horizontal keypoint based object detector to the oriented object detection task. In particular, we fifirst detect the center keypoints of the objects, based on which we then regress the box boundary aware vectors (BBAVectors) to capture the oriented bounding boxes. The box boundary-aware vectors are distributed in the four quadrants of a Cartesian coordinate system for all arbitrarily oriented objects. To relieve the diffificulty of learning the vectors in the corner cases, we further classify the oriented bounding boxes into horizontal and rotational bounding boxes. In the experiment, we show that learning the box boundary-aware vectors is superior to directly predicting the width, height, and angle of an oriented bounding box, as adopted in the baseline method. Besides, the proposed method competes favorably with state-of-the-art methods. Code is available at https:// github.com/yijingru/BBAVectors-Oriented-Object-Detection.
目前已有的用于遥感图像的方法主要是基于anchor的两阶段检测器
这些方法(例如R²CNN,ROI Transformer,R²PN,R-DFPN,ICN)通常使用中心点、宽度、高度和角度来定义定向边界框(OBB)
目前,为了克服上述anchor的缺点,基于关键点检测的算法被开发出来,这些方法检测bbox(bounding box)的角点,然后通过比较这些点的嵌入距离或中心距离来对这些点进行分组,这种方法可以很好的提高检测性能, 但是有一个缺点:分组步骤非常的耗时。
本文贡献主要有:
如上图所示,该网络的是建立在U型架构上的,使用ResNet101 Conv1-5作为骨干网络,在主干网络的顶部,对特征图进行上采样,并输出一个比输入图像小4倍的特征图,在上采样过程中,通过跳层连接将深层与浅层结合起来,以共享高级语义信息和低级更精细的细节,即首先通过双线性插值将深层变到与浅层相同大小,上采样的特征图通过3×3卷积层进行细化,然后将细化后的特征图与浅层连接起来,然后是一个1×1的卷积层来细化通道级的特征。在潜在层中使用了批处理归一化(BN)和ReLU激活