改进卷积神经网络算法在水面漂浮垃圾检测中的应用

Application of improved convolutional neural network algorithm in detection of floating garbage in water

针对传统目标检测算法在检测水面漂浮垃圾时易受外界复杂环境影响而难以实现的问题,提出了一种基于改进卷积神经网络的水面漂浮垃圾检测算法。运用数据增强技术改善训练过程中因样本不足而导致的过拟合问题,并利用迁移学习的方法训练出水面漂浮垃圾目标检测模型。结果表明,与传统的ViBe背景建模算法相比,所提算法能对水面漂浮垃圾进行分类,并标记出垃圾位置,对垃圾的检测准确率高达93%,能完全克服水波、波光等外界干扰。

In order to solve the problem that the traditional target detection algorithm is easy to be affected by the complex environment,an improved convolutional neural network algorithm was proposed.Data enhancement technology was used in the algorithm to overcome the over-fitting problem caused by insufficient samples in the training process.At the same time,the migration learning method was used to train the surface floating garbage target detection model.The results show that the algorithm can accurately detect the types of floating garbage on the surface and mark the location of garbage,compared with the traditional ViBe algorithm.The detection accuracy of garbage is up to 93%,and it completely overcomes external disturbances,such as water waves and sunlight.