About LOGO-Net

Logo detection is a special type of object detection in computer vision, which can be useful for many applications. Here we introduce a new logo image database for brand recognition called LOGO-Net, with a total of 100 brands, 160 logo categories, 73414 images, and 130,608 logo objects annotated with a bounding box and category label. Using Deep Region-based convolutional neural network (DRCN), we learn logo object detectors and apply them for brand recognition tasks, and establish new state of the art on logo detection benchmarks on real-world product images. Here we provide the LOGO-Net database and the trained collections of R-CNNs for academic research and education purposes.

News

• 5 Nov, 2015: The technical report has been submitted to arXiv.

• Datasets under preparation for public release of benchmark evaluations (it may take a while)

Download

• Logos-18: A logo image dataset which contains 8,460 images, 16,043 annotated logo objects from 18 logo categories.

• Logos-160: A logo image dataset which contains 73,414 images, 130,608 annotated logo objects from 160 logo categories..

• Logo detection demo: Input a picture and see how our method detects the logo.

• Evaluation code: Sample matlab code for performance evaluation.

References

Please cite the following paper if you use the database or the models

• S.C.H. HOI, X. WU, H. LIU, Y. WU, H. WANG, H. XUE, Q. WU. “LOGO-Net: Large-scale Deep Logo Detection and Brand Recognition with Deep Region-based Convolutional Networks.” ArXiv, 2015.
  [ PDF ] | [ Supplementary Materials ]

@inproceedings{LOGO-Net-15,
  title     = {Large-scale Deep Logo Detection and Brand Recognition with Deep Region-based Convolutional Networks},
  author    = {Steven C.H. HOI and Xiongwei WU and Hantang LIU and Yue Wu and Huiqiong Wang and Hui Xue and Qiang Wu},
  year      = {2015},
  booktitle = {arXiv}
}

Acknowledgement

This project was collaborated with Alibaba Group. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Alibaba Group. The annotations can be used under the License. The copyright of all the product images belongs to the image owners.

People and Collaborators

Principal Investigator: Dr Steven HOI
For any question about the LOGO-net, please email: LOGOnetdata@gmail.com For research collaboration, please email Dr Hoi: chhoi@smu.edu.sg
Team Members and Collaborators:
-Xiongwei WU (SMU), Hantang Liu (SMU), Yue Wu (SMU),
-Huiqiong Wang (Alibaba Group), Hui Xue (Alibaba Group), Qiang WU (Alibaba Group).