How do I classify ripeness of pineapple on real- time ? I have developed ripeness classification based on deep neural network but I don't know how to make it real time. Please help me

I have developed ripeness classification based on deep neural network but I don't know how to make it real time. Please help me.

[Ameetha]

Can You send the code if you were successful in implementing this?

Answer 1

[Image Analyst]

You have to develop an image analysis routine and put it on a portable laptop or a smartphone.

I'm not sure what defines ripe or not ripe - is it the size or color? If size, you might have to know your spatial calibration (for example have a scale or something of known size in the image). If color, the image sensor should probably have hyperspectral capability, though if you're lucky you might be able to get by with RGB.

Then you need to take lots of photos and label them as "ripe enough" or "not ripe enough". Then train your network by feeding all those labeled images through your network. Then deploy the network to a portable computing device with hyperspectral imager.

No, I don't have code for all this - it would take quite a while to develop. Since your need is urgent, I think you're totally out of luck. Even hiring a consultant to do it for you would take weeks at best. It's definitely no one hour job!

[Wyot]

Thank you very much

Answer 2

[Ameetha]

I have a similar question. I have trained a model to classify strawberries based on thier quality into GOOD and BAD quality. But I am not able to implement it in real time. pls help. Using the split data features of the mutually exclusive test dataset, I get very good accuracy. But I dont get correct classification result when extenal images are used.Can you help..

[Image Analyst]

Test data will always have lower accuracy than training data. If it's substantially lower then your training images are not representative of your test images. You should add images to your training set that are more similar to your test set.