AbdulazizAlshamsi/VLM_Dataset_classification

视觉过滤数据集是一个高质量、标记过的图像数据集，旨在为计算机视觉模型过滤噪声图像数据提供基准测试，尤其适用于视觉语言模型（VLMs）的预训练和数据集整理。数据集包含6个图像类别：图表、图表、几何图形、医学图像、包含文本的图像和随机图像，适用于训练和评估用于自动过滤大规模抓取数据集中的相关图像的分类模型。

The Vision Filtering Dataset is a high-quality, labeled image dataset designed to benchmark computer vision models for filtering noisy image data—especially relevant for pretraining and curating datasets for vision-language models (VLMs). The dataset contains 6 image categories: charts, diagrams, geometry, medical, OCR, and random, intended for training and evaluating classification models to automatically filter relevant images from large-scale scraped datasets.