hanif-ali / oocam-ml-interface

Repository for the OOCAM ML Interface

For using the current interface, please follow these steps -

1. Clone the repo to a suitable location:
   git clone https://github.com/shark-trek/OOCAM-ml-interface.git
2. Install dependencies :
   (i) Navigate to the directory where the repo has been cloned
   (ii) Double click the installer.bat file
3. Open Web Interface:
   (i) Open the 'flaskInterface.py' file
   (ii) Open the 'run.html' file
4. To train model:
   (i) In the first text field (File:), select the folder where all the images for training are located with the file selector.
   File name format: <class_name><image_number>.<jpg|png|jpeg>, eg: DotManta34.jpg
   (ii) In the second text field (Split:), enter the split value : a float value between 0 and 1
   0.8 would mean 80% of the images are for train, 20% for test
   (iii) In the third text field (Epoch:), enter number of epochs: no. of complete presentations of dataset to be learned by machine
   (iv) Click Train and give it a little time to process
   The model (model.h5) and the labels file (labels.dat) will be generated and placed into the 'output' folder
   The maximum validation accuracy achieved shall be shown on a new webpage
5. To predict an image:
   (i) In the first text field (Model:), enter the location of the model.h5 file
   (ii) In the second text field (Images:), enter the location of the folder containing the image files that needs to be predicted
   (iii) In the third text field (Labels:), enter the location of the labels.dat file
   (iv) The images would be sorted and stored in the 'predictions' folder in the main directory of the repo