In this case study, I combine several subcategories of data not based on item color but based on item type. These types will be classified into types of clothing and shoes. The result of merging these subcategories is that there are several main categories which are divided into 5 categories, namely Dress, Pants, Shirt, Shoes, Shorts. I'm trying to learn and build a deep learning approach with CNN Architecture using the pre-trained model "InceptionResNetV2" variant and the RMSprob optimizer to fit the data and then used to predict the image data. And in the last part, the model will be saved in TF-LITE format for development purposes (such as mobile applications).
Description
- Context: This dataset have been created by the author to practice multi-label classification.
- Content: The dataset consist of 11385 images.
For more information about the pretrained model, click link below: https://keras.io/api/applications/