Department of Computer Technology and Computation, University of Alicante, 03690 San Vicente del Raspeig, Spain.
Faculty of Science, University of Tuebingen, 72076 Tuebingen, Germany.
Sensors (Basel). 2023 Apr 2;23(7):3691. doi: 10.3390/s23073691.
Certain fields present significant challenges when attempting to train complex Deep Learning architectures, particularly when the available datasets are limited and imbalanced. Real-time object detection in maritime environments using aerial images is a notable example. Although SeaDronesSee is the most extensive and complete dataset for this task, it suffers from significant class imbalance. To address this issue, we present POSEIDON, a data augmentation tool specifically designed for object detection datasets. Our approach generates new training samples by combining objects and samples from the original training set while utilizing the image metadata to make informed decisions. We evaluate our method using YOLOv5 and YOLOv8 and demonstrate its superiority over other balancing techniques, such as error weighting, by an overall improvement of 2.33% and 4.6%, respectively.
某些领域在尝试训练复杂的深度学习架构时会面临重大挑战，特别是在可用数据集有限且不平衡的情况下。使用航空图像进行实时海上环境目标检测就是一个显著的例子。尽管 SeaDronesSee 是针对该任务最广泛和完整的数据集，但它存在严重的类不平衡问题。为了解决这个问题，我们提出了 POSEIDON，这是一种专门为目标检测数据集设计的数据增强工具。我们的方法通过结合原始训练集中的对象和样本，同时利用图像元数据做出明智的决策，生成新的训练样本。我们使用 YOLOv5 和 YOLOv8 来评估我们的方法，并通过分别提高 2.33%和 4.6%，证明了我们的方法优于其他平衡技术，如错误加权。
