Create my own datasets in tensorflow


I want to use my own images for training! I want to know how to:

  1. Label the data using tensorflow or other tools in the course
  2. Load the data for further training

Please take me to the thread or give me general advices, a tutorial would be great!

Thank you in advance,

There are plenty of tools available, like LabelImg, for labeling the images. Just Google it. You can also watch some guide on YouTube about LabelImg. Or you may check LandingAi tool.
Once your data is ready, loading it is just a snap.


Hello sir! I’m doing a much easier task here, I already have the images, I just need to make a “map” to assign the true values to the images. But currently I have no clue of doing it. Should I create a dictionary first? And pass the values in the learning process? I need some inspirations to streamline the process.

Simply put, I want to perform the same task as the Deep Learning Course 4 W2A2:

  1. Put images of the same class in the same folder
  2. Let the program learn the class in the folder.
    Upon checking the codes, I’m not sure if these functions do the tasks automatically
    image_dataset_from_directory( )

The best way to find the answer is to try. You may try this and post us here, whether this works or not. So, change the images of alpaca and non alpaca with your data and explore.

1 Like