Logo Classification #6
Description
I tried this challege using SIFT feature extractor and SVM and had for fine accuracy given the complexity of the dataset.
But I had to shift Conv net using TensorFlow and to use that had augment the dataset to increase number of images, and also reduced the classes to 10.
But for such a dataset I have to use atleast 5 layers of convolutions which is taking lots of time and also my system hangs in between.
I also tried Google Cloud Platform but they donot offer Nvidia GPU in India.
So, I would like to request that you allow only the code to be submitted for this task at least.
And reducing the epochs also doesn't help