OceanCV/PlasticInWater

This model was trained on images of different types of plastic placed inside a tank connected to a 1080p webcam, located in the Ocean Technology Center at the University of Washington. The goal of this project is to enhance the detection and classification of various types of plastic debris commonly found in marine environments, providing valuable tools for environmental monitoring and research. --- license: MIT --- ### Dataset Details The dataset consists of **4,511 images**, capturing diverse types of plastic materials. Plastics were annotated using the **YOLOv8 format** to facilitate accurate object detection. The types of plastic materials included are: - Black Plastic Cap - Blue Nitrile Glove - Blue Plastic Cap - Brown Multilayer Plastic - Green Plastic Cap - Orange Plastic Cap - Plastic Bottle - Purple Insulation Foam - Purple Multilayer Plastic Bag - Red-Orange BOPP Bag - Red Cap - Red Netting - Red Plastic Straw - Yellow Foam - Yellow Rope ### Pre-processing Steps The following pre-processing steps were applied to each image: 1. **Auto-orientation of pixel data** (with EXIF-orientation stripping) to ensure uniformity across all images. 2. **Resize to 640x360** (stretch) to standardize image dimensions for model training. No additional image augmentation techniques were applied to this dataset. ### Model Configuration - **Number of Classes (nc):** 15 - **Class Names:** See the list above for detailed class names. This model aims to support ongoing research in marine pollution by identifying plastic debris types effectively, helping researchers analyze patterns and develop mitigation strategies.本模型的训练数据集采集自华盛顿大学海洋技术中心（Ocean Technology Center, University of Washington）内的实验水槽，水槽搭载1080p网络摄像头，拍摄置于水槽内的各类塑料样本图像。本项目旨在提升海洋环境中常见各类塑料废弃物的检测与分类能力，为环境监测与科研工作提供实用工具。 --- 许可证：MIT --- ### 数据集详情 本数据集共包含**4511张图像**，涵盖多种不同类型的塑料样本。所有塑料样本均采用**YOLOv8格式**进行标注，以保障目标检测任务的精准性。 本数据集涵盖的塑料类型如下： - 黑色塑料瓶盖 - 蓝色丁腈手套 - 蓝色塑料瓶盖 - 棕色多层塑料 - 绿色塑料瓶盖 - 橙色塑料瓶盖 - 塑料瓶 - 紫色隔热泡沫 - 紫色多层塑料袋 - 红橙色双向拉伸聚丙烯（BOPP）袋 - 红色瓶盖 - 红色网材 - 红色塑料吸管 - 黄色泡沫 - 黄色绳索 ### 预处理步骤 本数据集对每张图像均执行以下预处理操作： 1. **像素数据自动校正**（移除EXIF方向元数据），以确保所有图像的归一化一致性； 2. **拉伸缩放至640×360分辨率**，统一模型训练所需的图像尺寸标准。 本数据集未使用额外的图像增强技术。 ### 模型配置 - **类别数量（nc）**：15 - **类别名称**：详见上述类别列表。 本模型旨在通过精准识别塑料废弃物种类，为海洋污染相关的持续研究提供支持，助力科研人员分析污染分布规律并制定减排治理策略。