Machine Learning - Final Assignment - Capstone Project
I have attached following files with the project. Folders
- Training folder Contains 10 training folders from n0 to n9
- Validation folders Contains 10 validation folders from n0 to n9
I couldn't create the folders and files due to huge size. Thus, I am providing the zip repository of the file. We need to unzip the folder and keep the files in the following folder structure:
-
./training/n0
-
./training/n1
-
...
-
./training/n9
-
./validation/n0
-
./validation/n1
-
...
-
./validation/n9
I have used 'Dog Breed' workspace to create this project. Details could be checked under folder 'Capstone_Project'.
Files
- Proposal.pdf Original proposal document
- Capstone - Project Report.pdf Project report
- Monkey_labels.csv File containing details of the input files
- Monkey_species.ipynb Python notebook containing python code with sample data run
- Monkey_species.html HTML version of python code with sample data run
- Monkey_species.pdf PDF version of python code with sample data run
I would like to call out various resources which influences my project
- Fine-Grained Categorization: https://vision.cornell.edu/se3/fine-grained-categorization/
- Fine-Grained Categorization: https://www.researchgate.net/publication/301452581_Croatian_Fish_Dataset_Fine-grained_classification_of_fish_species_in_their_natural_habitat
- https://www.kaggle.com/slothkong/10-monkey-species: There were various samples / recommendataions. They indeed help me think through my approach and how I go about the problem.
- Fig Ref: https://towardsdatascience.com/epoch-vs-iterations-vs-batch-size-4dfb9c7ce9c9
Additional References 5. Ref: https://isaacchanghau.github.io/post/loss_functions/ 6. Fig Source: https://www.cs.toronto.edu/~frossard/post/vgg16/ 7. Google for ResNet50 information