论文标题
动态比例训练对象检测
Dynamic Scale Training for Object Detection
论文作者
论文摘要
我们提出了动态量表训练范式（缩写为DST），以减轻对象检测中的比例变化挑战。以前的策略，例如图像金字塔，多尺度训练及其变体旨在准备标准不变数据以进行模型优化。但是，准备过程并不意识到以下优化过程，该过程限制了其处理量表变化的能力。相反，在我们的范式中，我们使用从优化过程中的反馈信息来动态指导数据准备。所提出的方法令人惊讶地简单，但获得了显着的收益（CoCo数据集的平均精度为2％+平均精度），表现优于先前的方法。实验结果证明了我们提出的DST方法对尺度变化处理的功效。它还可以推广到各种骨干，基准和其他具有挑战性的下游任务（例如实例分段）。它不会引入推理开销，并且可以作为免费的午餐，以供一般检测配置。此外，由于快速收敛，它还促进了有效的培训。代码和型号可在github.com/yukang2017/stitcher上找到。
We propose a Dynamic Scale Training paradigm (abbreviated as DST) to mitigate scale variation challenge in object detection. Previous strategies like image pyramid, multi-scale training, and their variants are aiming at preparing scale-invariant data for model optimization. However, the preparation procedure is unaware of the following optimization process that restricts their capability in handling the scale variation. Instead, in our paradigm, we use feedback information from the optimization process to dynamically guide the data preparation. The proposed method is surprisingly simple yet obtains significant gains (2%+ Average Precision on MS COCO dataset), outperforming previous methods. Experimental results demonstrate the efficacy of our proposed DST method towards scale variation handling. It could also generalize to various backbones, benchmarks, and other challenging downstream tasks like instance segmentation. It does not introduce inference overhead and could serve as a free lunch for general detection configurations. Besides, it also facilitates efficient training due to fast convergence. Code and models are available at github.com/yukang2017/Stitcher.
