I am doing a final year project about image processing in production lines (An image processing system consisting of cameras and a tracker to
detect defects and to verify correct position of labels, in a
production line)
We will be using Arduino to control the camera as it rotates to capture images of products in a production line for image processing. I want to know if Arduino can store those images (what kind , what type) and is Arduino capable of sending those images IN REAL-TIME to a computer that is installed with a machine learning tool (yolov8) so that it can study them for defects (1 defect to be specifically) ? we are planning on using Arduino MEGA
Arduino has an 8Bit processor. Production speed is very important in error detection. You should develop a project with high FPS. I think Raspberry Pi 5 would be suitable for this. Of course, lighting the environment is also very important.
No, but for additional help or suggestions, specify what kind and size of images you plan to collect. The more details you supply, the better chance of learning something useful.
Hmm, that brings back memories; it was my thesis in 1988.
We used a Motorola 68000 based computer systems with framegrabber and 512x512 8-bit camera.
I'm not sure what kind of environment your production line is. Ours was in the food and drinking industry and that can be extremely filthy. So make sure that whatever you design can be used in that environment.
The Unisys DP1800 high speed check reader/sorter was based on the 68000 series. In the 1990's they added image capture of the checks, front and back, while the checks were passing at 1800 per minute.
I am writing because the OP never mentioned the light source for his imaging. Unisys used zenon lights to flash for each image because the light was very close to being pure white. The correct light will have a great effect on the performance of the image being taken.