Fruit damage detection

asked 2016-07-07 12:00:54 -0600

SimonG gravatar image

updated 2018-01-01 11:15:24 -0600

Hi,

I am 16 years old and I am quite good at coding. My uncle is a fruit harvester and they have a really hard year this year because of the wheater. Because I am quite good at coding and I am getting into electronics (Arduino, Raspberry pi,..) I thought I could make a sorting system for him. It does not seem so hard. I am not quite good at image processing, the hard part for me. I am just getting into it, and now I am thinking about the best way of processing fruit damage. I am beginning with Cherry's, because it is the season for it right now :') I have taken a couple pictures which I will include.

I was thininking about filtering the red/black colors out and leaving brown colors. (Because most of the damage is turned brown) Than I just need to detect if there is some brown color within the cherry circle? Is that a good approach? Or should I do it in some other way? It would be nice if anyone has a good idea, that will give me a point to start :')

Thankyou! Here are some pictures, not everything is damaged. image description image description image description image description image description image description image description image description image description image description image description image description image description

edit retag flag offensive close merge delete

Comments

It does not seem so hard. Can I smack you on the head? :) Of course this is doable with machine vision and machine learning, but it won't be easy. Thats the exact reason why research groups and companies like ours are full time focussing on computer vision tasks. What I would do is take color histograms, texture features, ... and then use SVM to classify the feature vectors.

StevenPuttemans gravatar imageStevenPuttemans ( 2016-07-08 04:26:02 -0600 )edit