Image Classification - is it a cat or a dog?
The ultimate goal of this project is to create a system that can detect cats and dogs. While our goal is very specific (cats vs dogs), ImageClassifier can detect anything that is tangible with an adequate dataset.
Dataset: Dogs vs Cats
Description: Binary classification. Classify dogs and cats.
Training: 20,000 images (10,000 per class)
Validation: 5,000 images (2,500 per class)
Testing: 12,500 unlabeled images
Input Data Shape: 64x64x3
Convolutional Layer 32 filter Filter shape: 3x3
Activation Function: ReLu
Max Pooling Pool shape: 2x2
Convolutional Layer 32 filter Filter shape: 3x3
Activation Function: ReLu
Max Pooling Pool shape: 2x2
Classification:
Activation Function: ReLu
Dropout Rate: 0.5
Activation Function: Sigmoid