gogul004/metal_scrap_dataset

该数据集包含金属废料的合成图像，根据质量和可回收性分为三个等级。图像使用Stable Diffusion生成，并通过增强技术增加了多样性和鲁棒性，适用于机器学习任务。数据集包含约13,895张增强后的图像，分为三个类别：A级（5,085张图像，1,017张原始图像）- 干净、无锈金属；B级（5,570张图像，1,114张原始图像）- 轻微锈蚀、可回收金属；C级（3,240张图像，648张原始图像）- 不可回收、严重降解的废料。图像大小为224x224像素，格式为PNG。每张原始图像通过调整大小和多种增强技术（如随机翻转、旋转、颜色抖动、仿射变换等）生成了4个变体。数据集适用于金属废料分类、回收分析或计算机视觉研究的模型训练。

This dataset contains synthetic images of metal scraps, categorized into three grades based on quality and recyclability. The images were generated using Stable Diffusion and augmented to enhance variety and robustness for machine learning tasks. The dataset includes approximately 13,895 augmented images, divided into three categories: Grade A (5,085 images, 1,017 original) - Clean, rust-free metals; Grade B (5,570 images, 1,114 original) - Lightly rusted, recyclable metals; Grade C (3,240 images, 648 original) - Unrecyclable, heavily degraded scrap. Image size is 224x224 pixels, in PNG format. Each original image was resized and augmented with 4 variations using techniques like random flips, rotation, color jitter, affine transformations, etc. The dataset is intended for training models in metal scrap classification, recycling analysis, or computer vision research.