images.cv is a free catalog of labeled image datasets for computer vision. Browse, search, and download datasets for classification, detection, and segmentation — no signup required.

Researchers, students, ML engineers, and product teams that need ready-to-train data for POCs, course work, fine-tuning, or baseline models.

No. Every dataset on images.cv is free to download anonymously. If you want downloads to arrive by email or save your search history, you can create a free account.

Millions across thousands of categories. Use the search page to filter by keyword, class, and minimum image count.

Choose output image size, color mode (RGB / grayscale), and train/val/test split before you download. Datasets arrive as a ZIP with an ImageFolder-style layout ready for PyTorch, TensorFlow, or Keras.

Yes. Pick augmentations (rotate, flip, crop, blur, noise, color jitter, and more) and choose what percentage of the train split should be randomly augmented. A slider lets you set the ratio from 0% to 100%.

Images are aggregated from public web sources and community contributions. Datasets are intended for research and educational use; always verify the underlying license for your specific use case.

The download page shows a live size estimate based on your image size, color mode, and augmentation settings. Small datasets are a few MB; the largest reach several GB.

Yes — our B2B team delivers production-grade labeled datasets with QA, custom classes, and annotation formats (YOLO, COCO, segmentation masks). See For Business for scope and timeline.

images.cv is a free resource supported by voluntary donations (Buy Me a Coffee) and our B2B custom-dataset service. There are no paid consumer tiers for the public catalog.

Open any dataset and use the report button on the image thumbnail — our team reviews reports and removes problematic samples.