遥感图像有向目标检测是遥感图像解译中的一项基础任务，在许多领域有着广泛的应用。由于遥感图像目标尺度差异性大、方向任意且紧密排列，传统目标检测所使用的水平框无法准确的定位目标。因此，遥感图像有向目标检测成为目前遥感领域的研究热点。受益于深度学习的发展，遥感图像有向目标检测取得了突破性进展，但是大多数方法仅在检测头部加入角度预测参数，在训练过程中没有充分利用角度信息和语义信息。本文提出了一种多元信息监督的遥感图像有向目标检测方法。首先，在感兴趣区域提取阶段利用角度信息监督网络学习目标方向，从而使网络第一阶段生成更加贴近遥感图像目标的有向候选区域。其次，为了充分利用图像语义信息，本文在网络第二阶段增加语义分支，并使用图像语义标签进行监督学习。本文以Faster R-CNN OBB为基准，在DOTA数据集上验证所提方法的有效性。本文方法相比基准，平均精度（mAP）提升了2.8%，最终的检测精度（mAP）达到74.6%。

Objective: Oriented object detection is a basic task in high-resolution remote sensing image interpretation. Compared with general detectors, oriented detectors can locate instances with oriented bounding boxes, which are consistent with arbitrary-oriented ground truths in remote sensing images. Currently, oriented object detection has made great progress with the development of the Convolutional Neural Network (CNN). However, this task is still challenging because of the extreme variation of object scales and arbitrary orientations. Most oriented detectors are evolved from horizontal detectors. They first generate horizontal proposals using Region Proposal Network (RPN), then classify those proposals into different categories and transform them into oriented bounding boxes. Despite their success, those detectors just exploit the annotations at the end of the network and do not make full use of the angle and semantic information. Method: This work proposes an Angle-based Region Proposal Network (ARPN) which learns the angle of objects and generates oriented proposals. The structure of ARPN is the same as the RPN. However, for each proposal, instead of outputting four parameters for regression, ARPN generates five parameters which are the center (x, y), shape (w, h) and angle (t). In training, we first assign anchors with ground truths by the Intersection of Unions (IoUs). Then we directly supervise the ARPN with the shape and angle information of ground truths. In order to take the advantage of the semantic information, we also propose a semantic branch to output image semantic results. The semantic branch consists of two convolutional layers and is parallel with the detection head. We first assign objects to different scale levels according to their areas. Then, we create semantic labels in each scale and use them to supervise the semantic branch. With the semantic information supervision, the model will learn translation-variant features and improve accuracy. Besides, the outputs of the semantic branch indicate the objectness in each place, which can filter out false positives of final predictions. Results: We conduct comprehensive experiments on the DOTA dataset to validate the effectiveness of the proposed methods. In data preparation, we first crop original images into 1024×1024 patches with the stride of 824. Compared with baseline, the ARPN obtains 2.2% mAP gain and the semantic branch boosts 0.8% mAP. Finally, we combine both methods and achieve 74.64% mAP, which is competitive with other oriented object detectors. We visualize some results on the DOTA dataset. The results show that our method is much effective on small objects and densely packed objects. Conclusion: In this paper, we proposed ARPN and the semantic branch to utilize the multi-information in remote sensing images. The ARPN can directly generate oriented proposals which can better recall oriented objects. The semantic branch increases the translation-variant of the features. Experiments demonstrate the effectiveness of our method, which achieves 74.64% mAP on the DOTA dataset. In feature works, we will focus on the model efficiency and the inference speed.