Landscape Pictures
Image dataset composed of 4,319 real photos of various landscapes collected on Flickr. These images cover several themes: mountains, deserts, seas, seas, beaches, islands, and Japanese landscapes. Used in particular for artistic style transfer experiences such as CartoonGan.
4,319 JPG images, various categories (mountain, desert, desert, sea, beach, island, Japan)
CC0: Public Domain
Description
The dataset Landscape Pictures includes 4,319 photos from Flickr, classified into several natural categories such as mountain landscapes, deserts, seas, beaches, islands, and Japan. These images are of high quality and represent real scenes, providing a rich basis for computer vision and style transfer work.
What is this dataset for?
- Test art style transfer methods like CartoonGan
- Develop models for classifying or segmenting landscape images
- Train GaNs to generate realistic or stylized images
Can it be enriched or improved?
Yes, this dataset can be completed with other images from similar free archives or annotated more finely for the precise segmentation of landscapes (e.g. sky, vegetation, water). We can also experiment with different resolutions and augmenting transformations.
🔎 In summary
🧠 Recommended for
- Vision researchers
- GAN developers
- Digital artists
🔧 Compatible tools
- FastAI
- PyTorch
- TensorFlow
- CartoonGan
- OpenCV
💡 Tip
Use categories for targeted style transfer experiences by landscape type.
Frequently Asked Questions
Does this dataset contain annotated images for segmentation?
No, annotations are limited to global categories without pixel by pixel segmentation.
Are the images royalty-free for commercial use?
Yes, the CC0 license allows unrestricted commercial use.
Can this dataset be used to train image generation GANs?
Yes, this dataset is perfectly suited for GaNs and style transfer experiences.
