Jetson Nano and Image Inspection Applications

Autonomous Machines Jetson & Embedded Systems Jetson Nano
1

Hi … I tried to look around and see if this topic is covered anywhere and could not find examples or a knowledge-base to help …

As an example, I want to use the Nano to find identical pencils based on

  1. Size of its eraser
  2. Length of the pencil
  3. Width of the pencil
  4. If the pencil has a point on it.
  5. Color of the pencil

At first I thought I would need to create a network of all the possible options. But that would take a lot of time to create. So, can I give it at the start of a sorting run the image of the pencil I want to find and then it compares that master against the thousands of images of different pencils I have, giving a green light to those that match my 5 matching criteria?

2

Hi,

This should be an open question.
Here are some suggestions from our side:

1. Do the match in a DNN way
To apply the five feature mentioned above, you will need several DNNs first:

  • A pencil detector
  • A point classifier
  • A color classifier

Then implement an algorithm to combine all the information above and output the final result.

2. Do it in a traditional computer vision way.
This looks like a standard template matching problem to me.
It’s worthy to check if the implementation in OpenCV is sufficient for your or not:
https://docs.opencv.org/master/d4/dc6/tutorial_py_template_matching.html

Thanks.

3

Thanks … this gives me an excellent next step …