Key Laboratory of Dynamics and Control of Flight Vehicle, Ministry of Education, Beijing Institute of Technology, Beijing 100081, China.
Sensors (Basel). 2021 Feb 4;21(4):1066. doi: 10.3390/s21041066.
At present, the one-stage detector based on the lightweight model can achieve real-time speed, but the detection performance is challenging. To enhance the discriminability and robustness of the model extraction features and improve the detector's detection performance for small objects, we propose two modules in this work. First, we propose a receptive field enhancement method, referred to as adaptive receptive field fusion (ARFF). It enhances the model's feature representation ability by adaptively learning the fusion weights of different receptive field branches in the receptive field module. Then, we propose an enhanced up-sampling (EU) module to reduce the information loss caused by up-sampling on the feature map. Finally, we assemble ARFF and EU modules on top of YOLO v3 to build a real-time, high-precision and lightweight object detection system referred to as the ARFF-EU network. We achieve a state-of-the-art speed and accuracy trade-off on both the Pascal VOC and MS COCO data sets, reporting 83.6% AP at 37.5 FPS and 42.5% AP at 33.7 FPS, respectively. The experimental results show that our proposed ARFF and EU modules improve the detection performance of the ARFF-EU network and achieve the development of advanced, very deep detectors while maintaining real-time speed.
目前，基于轻量级模型的单阶段探测器可以实现实时速度，但检测性能具有挑战性。为了提高模型提取特征的辨别力和鲁棒性，并提高探测器对小物体的检测性能，我们在这项工作中提出了两个模块。首先，我们提出了一种称为自适应感受野融合（ARFF）的感受野增强方法。它通过自适应学习感受野模块中不同感受野分支的融合权重，增强模型的特征表示能力。然后，我们提出了一个增强上采样（EU）模块，以减少特征图上上采样引起的信息丢失。最后，我们在 YOLO v3 上组装 ARFF 和 EU 模块，构建一个实时、高精度、轻量级的目标检测系统，称为 ARFF-EU 网络。我们在 Pascal VOC 和 MS COCO 数据集上实现了速度和精度的最新水平，分别在 37.5 FPS 下达到 83.6%的 AP 和在 33.7 FPS 下达到 42.5%的 AP。实验结果表明，我们提出的 ARFF 和 EU 模块提高了 ARFF-EU 网络的检测性能，并在保持实时速度的同时实现了先进的、非常深的探测器的发展。
