[论文解读] Deep Neural Networks for Marine Debris Detection in Sonar Images

Matías Valdenegro-Toro|arXiv (Cornell University)|May 13, 2019

Advanced Neural Network Applications参考文献 134被引用 26

一句话总结

本文提出一种基于卷积神经网络（CNNs）的深度学习方法，用于在前视声呐（FLS）图像中检测海洋垃圾，利用在水箱中采集的2,069张FLS图像构建的自定义数据集。该方法在匹配和检测建议任务中显著优于当前最先进技术，尤其在样本复杂度和目标尺寸敏感性方面表现突出。

ABSTRACT

Garbage and waste disposal is one of the biggest challenges currently faced by mankind. Proper waste disposal and recycling is a must in any sustainable community, and in many coastal areas there is significant water pollution in the form of floating or submerged garbage. This is called marine debris. Submerged marine debris threatens marine life, and for shallow coastal areas, it can also threaten fishing vessels [Iñiguez et al. 2016, Renewable and Sustainable Energy Reviews]. Submerged marine debris typically stays in the environment for a long time (20+ years), and consists of materials that can be recycled, such as metals, plastics, glass, etc. Many of these items should not be disposed in water bodies as this has a negative effect in the environment and human health. This thesis performs a comprehensive evaluation on the use of DNNs for the problem of marine debris detection in FLS images, as well as related problems such as image classification, matching, and detection proposals. We do this in a dataset of 2069 FLS images that we captured with an ARIS Explorer 3000 sensor on marine debris objects lying in the floor of a small water tank. The objects we used to produce this dataset contain typical household marine debris and distractor marine objects (tires, hooks, valves, etc), divided in 10 classes plus a background class. Our results show that for the evaluated tasks, DNNs are a superior technique than the corresponding state of the art. There are large gains particularly for the matching and detection proposal tasks. We also study the effect of sample complexity and object size in many tasks, which is valuable information for practitioners. We expect that our results will advance the objective of using Autonomous Underwater Vehicles to automatically survey, detect and collect marine debris from underwater environments.

研究动机与目标

• 开发并评估深度神经网络（DNNs），用于自动检测水下声呐图像中的海洋垃圾。
• 解决利用真实水下环境中声呐数据检测沉没海洋垃圾（通常由可回收材料构成）的挑战。
• 在受控水箱环境中，基于传感器在真实环境中的限制，构建一个包含2,069张FLS图像的综合性基准数据集，涵盖10类垃圾和背景类别。
• 研究样本复杂度和目标尺寸对自主水下航行器（AUVs）实际部署中检测性能的影响。
• 推进配备基于DNN视觉系统的AUV，实现海洋垃圾调查与收集的自动化。

提出的方法

• 使用在水箱中通过ARIS Explorer 3000传感器采集的2,069张前视声呐（FLS）图像构建自定义数据集，模拟真实海底环境。
• 设计并训练深度神经网络（DNNs），包括卷积神经网络（CNNs），用于多个任务：图像分类、图像匹配和目标检测建议生成。
• 采用标准深度学习组件：Conv( f, w×h ) 表示卷积层，MaxPool( w×h ) 表示最大池化层，AvgPool 表示全局平均池化，FC( n ) 表示全连接层。
• 使用小批量梯度下降（MGD）进行模型训练，学习率 α，批量大小 B，最多训练 M 个周期，基于验证集（Vl）进行优化，并在保留的测试集（Ts）上进行测试。
• 使用Keras、Theano和scikit-learn进行实现，通过Robotarium集群进行高性能GPU计算，训练超过1,000个神经网络。
• 在检测建议和匹配等任务中评估模型性能，分析目标尺寸和训练样本数量的敏感性。

实验结果

研究问题

• RQ1 深度神经网络能否在前视声呐图像中检测海洋垃圾方面超越当前最先进方法？
• RQ2 样本复杂度（训练样本数量）如何影响DNN在海洋垃圾检测任务中的性能？
• RQ3 目标尺寸对声呐图像中检测准确率和建议质量有何影响？
• RQ4 DNN在生成准确检测建议和匹配相似垃圾物体方面效果如何？
• RQ5 受控水箱数据集在多大程度上能泛化到真实世界水下垃圾检测场景？

主要发现

• DNN在FLS图像中检测海洋垃圾方面显著优于当前最先进方法，尤其在检测建议和图像匹配任务中表现突出。
• 所提出的DNN框架在检测建议生成方面取得显著性能提升，显示出在自主水下航行器（AUV）系统中集成的强潜力。
• 样本复杂度被证实为关键因素，随着训练数据量增加，性能显著提升，尤其对小型垃圾物体更为明显。
• 目标尺寸对检测准确率有明显影响，小型垃圾物体更难检测，凸显了对尺寸感知模型设计的需求。
• 在分类、匹配和建议任务的综合评估中，DNN在水下垃圾检测中展现出鲁棒性和多功能性。
• 包含2,069张FLS图像（涵盖10类垃圾和背景）的自定义数据集，为声呐基海洋垃圾检测的未来研究提供了宝贵基准。