How to ask my trained model?

General Discussion
,
#1

Hello people,

I’m pretty new to TensorFlow and I need some advice !

First, I would like to train an image classification model with my own images. Then, I would like to be able to serve that model and make requests to it so as I can provide it an image and it would tell me if he recognizes something and with which accuracy.

What would be the perfect model ? And how could I send an image to this model so as it would send me what it have recognized ?

I don’t know if there’s already a tutorial explaining how to do it, I would love the link if there’s one.

Thank you very much !

#3

You might start with a Keras Image Classification Tutorial YouTube tutorial. It will take you through the concepts and training and testing of an actual Keras model for image classification.

1 Like
#4

Thank you @rcauvin ! Very interesting video ! What can I watch to solve the second part of my problem?

#5

I’m not sure if this is what you want, but you have to first make your image into a numpy array or a tensor, then you can call model.predict on your image.

ex.

image = tf.zeros((10, 10)) # creates a 10x10 blank image
model.predict(image) # model accepts image(make sure input shape of model is same dimension as image)