Logo Classifier is using Pytorch model to classify logos The supported logos are Adidas, Aldi, Apple, Becks, BMW, Carlsberg, Chimay, Coca-Cola, Corona, DHL, Erdinger, Esso, Fedex, Ferrari, Ford, Foster's, Google, Guiness, Heineken, HP, Milka, Nvidia, Paulaner, Pepsi, Ritter Sport, Shell, Singha, Starbucks, Stella Artois, Texaco, Tsingtao and UPS or No Logo if there is no logo exists.

FlickrLogos32 Collected logos of 32 different logo brands from Flickr. All logos have an approximately planar surface.

Pretrained resnet152 was used as a starting point. 0.8 of the data is used in training, 0.1 for validation and 0.1 for testing Logo_Classifier.ipynb shows the progress of the training and how we did it.

To use our application you just need to enter to the web app and upload your image file and it will tell you the prediction and confidence. The web app is deployed on Heroku.

Because of memory imitations Heroku app can be down, in this case you can run the app locally.

• Clone the repo git clone https://github.com/DavidAshraf/Logo-Classifier

• install dependencies, go to "web app" directory and run this command from pip3 install -r requirements.txt

• download the model in the same directory "web app"

• run the app python3 logo-classifier.py

We would like to thank Christian Eggert for providing us with the dataset

MIT