Logo-2K+ covers a diverse range of logo classes from real-world logo images, containing 167,140 images with 10 root categories and 2,341 categories.
For more information please read Logo-2K+: A Large-Scale Logo Dataset for Scalable Logo Classification.
Annotations consist of simple image labels. Labels are stored in .txt files that contain the image paths and are part of the paths.
You can download the dataset here.
The official DRNA-Net model can be consulted here.
No official benchmarks have been provided for this dataset.
No associated challenges have been found for this dataset.
Dataset licenced under a Non-Commercial licence.