Keras based CNN models for classification related problems .
The Dataset for both Image Classification problems from Kaggle :
Hydra : https://www.kaggle.com/c/invasive-species-monitoring
Plant Seedling Classification : https://www.kaggle.com/c/plant-seedlings-classification
The approach for both problems was to first Augment the data and then feed a pretrained network to extract features . These extracted features were then trained on a Neural Network to gain final results . I also tried using ensembling (SVM and Deep Logistic on the extracted features) but a neural network performed better .
Hydra :
The problem statement was to classify invasive species of Hydra and non-invasive species . I have used transfer learning (a pretrained VGG model) and Data Augmentation to get results .
Plant Seedling Classification :
The problem statement was to classify different plant seedlings into specific types . A Resnet model is used here to get results instead of A VGG .