Identify objects moving on a conveyor belt using Opencv with Python 

thank you, I wasn't aware of that, I've just noticed it now rewatching my video. I guess I pronunced it wrong so many times that it just came natural to me that way :)

@@pysource-com don't worry too much about it. It is the content we are after not the form. We all have automatic patches in our brains for these things:-) Great video, as always!! Thanks

​ @Pysource Can you explain this part, or can you make another video for this.

@@jayanthkumar8927 you would need a machine learning model probably or you could decide based on the size

The main purpose of this video was to focus just on the computer vision aspect of the project. The hardware part is actually the simplest one, and there is some reference about it on the blog post (link below in the description). Mostly the python code is connected to a Relay that turns on/off the motor.

Ok, thanks, a lot, I have another issue with the Python Editor that you are using namely 'Thonny', my cv2 is not working, and giving me the error as below:- import cv2 ModuleNotFoundError: No module named 'cv2' Even, I have installed cv2 packages from Tools--> Manage Packages, Can you help me what to do, please? Thanks, in Advance. best regards Gul

thanks. unfortunately this is not available for sale, it's a small steel conveyor made by a blacksmith

Yes it matters. There are specific pins to use for the servo motors. Personally I used a servo hat which made this process much simpler to install. I recommend to check servo hat for servo motors by adafruit

Have you been able to figure out the conveyor_lib ?

​@@mariammoufaddal5651 conveyor_lib is used by a file_handling technique which is basically a name of python file which is used to link ResberryPi signals with "main.py" file.As u can see the "import conveyor_lib" in third line of "conveyor_lib.py" file in which they were calling "conveyor_lib" from the conveyor_lib.py file. These are basically two files "conveyor_lib.py" and "main.py" which u will get by registering to the blog of this tutorial which they have mentioned in the description.

@@abubakarsaleem5167 thank you

Hi, unfortunately I decided to suspend the series because It is going to take too long to complete it, instead I'm focusing on other projects for the videos.

Considering in this project we compare the nuts by their size, probably in your case it won't be the best approach. You should find a way to distinguish them. For example, if you have cuboid objects and they need to be all with the same orientation, parallel to the belt, you could check the coordinates of each single object and compare it with the borders of the belt, if they're parallel or not.

Thanks a lot for your reply !!:)

Hi Pablo, I suggest to use cv2.GaussianBlur() as first step. Gaussianblur is better than just blur because it remove noise without removing details of the object, while the cv2.blur is less precise. Then cv2.erode(). and you're good to go.

@@pysource-com Thank you so much!

Yes it's reliable, it's just important to add the heatsinks so it doesn't get too hot and you can leave it always on.

@@pysource-com okay sir tq so much. If u post videos like measuring dimentions of a part and detecting cracks in future it will be very useful.

That lib is provided by the servo hat you're using. It depends on the brand! I recommend to get the adafruit servo hat because is the most user friendly to implement

just use contours, _ = cv2.findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE) you wont get the error

@@121horaa thanks

i was got the same error, but its solved with this code contours,hierachy = cv2. findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

I haven't personally tried raspberry pi 0, so I can't guarantee it will work. If you're able to run Opencv thogh, most certainly it's going to work, as the detection is very light. I did this project on a raspberry pi 3 or 4 (i don't remember which one exactly)

Nope, this one is basic opencv with python. For more complex detection you would need deep learning (whether with tensorflow, pytorch or other frameworks)

@@pysource-com okay thanks, but once again, what type measurement on this program? like mm, cm or else?

Yes, with a more powerful device than the raspberry pi you can achieve that results. But it depends also on how you want to detect them and if artificial intelligence needs to be involved

so thank you bro.

I used a simple webcam, the logitech c922pro. For high velocity is recommended to use industrial cameras with high FPS, and definitely faster devices than the raspberry pi

I got the same error message. Did you solve it?

replace the line contours line withe this line and it should work. dont use under line comma at start contours, hierarchy = cv2.findContours(threshold, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)