Hi all,
This code base is my attempt to give basic but enough detailed tutorial for beginners on image classification using keras in python.
-
I used here the basic set of libraries and predict 10 categories of image.
-
Implemented a simple CNN Model and used 10 images per category (10x10 = 100 rows of input) to Train and Test the model
-
Model performed >70% accuracy which is quite good for this small set of data.
-
You can add / update any kind of images and labels and Train the model to meet your needs.
-
You can find inline comments to understand the code easily and modify as per your requirement.
https://medium.com/@sivarajng/simplest-image-classification-in-keras-python-tensorflow-ea956b12bb42
# True : To Re-Train Model with New set of Images and Labels
# False : To Load existing Trained Model and simply predict
retrain = False
if(retrain):
imageClassification.start(retrain=True)
else:
imageClassification.start(retrain=False)