As a fundamental vision-based task in the remote sensing field, object detection has achieved significant progress. However, remote sensing object detection is still an urgent challenge owing to dense distribution, large aspect ratios, and arbitrary orientations. To address this issue, we develop an end-to-end dual align single-stage rotation detector (DA-Net) consisting of two main components: a rotation feature selection (RFS) module and a rotation feature align (RFA) module. Specifically, the RFS module can empower neurons with the capability of adjusting receptive fields, which achieves the first stage of feature alignment on the image level. Furthermore, the RFA module is employed to adaptively align the feature based on the size, shapes, and orientations of its corresponding anchors, realizing the second stage of instance-level feature alignment. Extensive experiments have shown that our DA-Net can significantly improve remote sensing detection performance against several state-of-the-art algorithms on two benchmark datasets.