Machine Learning Image Classifier Designed to Help Journalists and Photographers
Check out our final project here.
Our final project is a fully working ranking system that takes user-uploaded images and gives them a score. Then, the user is allowed to view and sort the images they've uploaded, and select any or all photos that they'd like to redownload.
This website design is based off of a bootstrap template. Server side code is handled via php, which interacts with a mySQL database. We're hosting our system on an Amazon EC2 instance, which again can be found here. The image classifier that we use to score and rank the images is the idealo image classifier, which we are also hosting on the EC2 server. This classifier has not been trained on our data set, and is simply the "out-of-the-box" model.
We elected to use Amazon Web Services to deploy our project. We selected the Amazon Linux AMI for our server as it already came with MySQL, Docker, and the appropriate Python libraries already installed. Should you chose to deploy our project on a different Linux environment, you will need to install Docker, MySQL and at minimum Python 2.7. You will also need to install Exiftool by Phil Harvey. There is not a repository to fetch Exiftool, so you will have to acquire it using wget to retrieve the needed packets; enter the following commands
wget http://owl.phy.queensu.ca/~phil/exiftool/Image-ExifTool-11.38.tar.gz
tar -xzf Image-Exiftool-11.38.tar.gz
cd Exiftool-11.38
sudo cp -r exiftool lib /usr/local/bin
Steps