Hello,
I am newbie to TensorFlow. I apologize I wasn’t able to search the forum for this question. I want to be develop a quick application (Python) to input an image and detect certain aspects (features) from the image using TensorFlow. Is this possible? I found this tutorial.
Can I use that for what I want to do?
Any help is much appreciated.
Mustafa
Hi @Mustafa_Cayci, Welcome to the TensorFlow Forum!
Yes, TensorFlow can be used to extract features from images. To get started, Please follow this official tutorial 1, tutorial 2, tutorial 3. Thank you!
Hello @Divya_Sree_Kayyuri Thank you so much for getting back to me to my question above. I am going to go through the tutorials you mentioned. May I come back to you if I have further questions?
Hi @Mustafa_Cayci, If you have any further questions or encounter any issues, we’d be happy to assist you.
Hi @Divya_Sree_Kayyuri I was reading this blog here. Does TensorFlow accomplish all three layers describe in the blog?
Mustafa
Hi @Mustafa_Cayci, Yes, We can implement all these layers in Tensorflow.
Hello @Divya_Sree_Kayyuri In our application implementation, the drone is configured to capture images at various stages during each flight along a predetermined path. It autonomously takes pictures, resizes them, and processes them. We want to fly over an area of crop field. We want to train the Tensor Flow for crop health by healthy versus diseased. In order to train the TensorFlow, do I need to supply a same size crop field with one healthy condition and the other one diseased condition? Can it be a small image of a plant that is healthy versus diseased? For example, if the crop field area is 10 acres, should the training data size be the same size 10 acres? Can the plant be same as the crop being analyzed?
Mustafa
Hello @Mustafa_Cayci , TensorFlow models for image classification are trained on small, individual images that represent each class. You do not need to supply a same-size, 10-acre crop field for both the healthy and diseased conditions.
Hello @Divya_Sree_Kayyuri Please refer to a diseased soybean image. Should the images be this close up in order to train TensorFlow? Because we fly drones at an altitude 20 - 100 meters. Clearly images at that altitude will not be this close up.
Hello @Divya_Sree_Kayyuri
I am following up on my previous query. Unfortunately, I need this information before our mission to ensure we capture the correct resolution imagery. Is there a premium support channel that provides more timely responses?
Thank you for your assistance.
Best regards,
Mustafa
