Machine Learning / Computer Vision Part 2

• By Transfer Learning With ResNet50 Pre-Trained Model Using Keras and Tensorflow 2.

SETUP

1. I specifically made this code for Kaggle Kernel (notebook). You can try to import this to Google Colab or use Jupyter Notebook on you local. Let me know if you got problems.
2. You have to gather (or produce) your own image dataset. There are lots available online as open data.
3. For your dataset, make three folders and categorize them as: train, valid and test. I tested at least 100 train images and 2o for validation images. In some cases, it might not be the same and most likely you will need more train images. It's up to you to tweak the numbers and ratio.
4. Once you run the code, it automatically trains and produces the .h5 file which actually is the "model".

CONTENT

I. File: roboteur-resnet50-model.ipynb

• Transfer learning / fine tuning code for training model with custom dataset and produces .h5 file as model.
• If you want to learn about the concepts, just message me via instagram or facebook.

II. File: roboteur-resnet50-inference.ipynb

• Codes for inferencing models and seeing the results.