CASE STUDY · COMPUTER VISION
Counting fruit on the line with colour and depth vision.
The client needed an AI solution that uses synchronized colour and depth images to detect and count fruit automatically on a production line.
PROJECT OVERVIEW
- INDUSTRY
- Food production
- PROJECT TYPE
- Computer vision detection system
- CAPABILITIES
- Computer Vision, Colour and Depth Imaging, Segmentation, Object Counting
01 · THE CHALLENGE
Counting by eye does not scale.
On a production line, fruit needs to be identified and counted quickly and consistently.
Doing this by eye is slow and repetitive, and apples that touch or overlap are easy to miss or count twice.
The goal was a system that could detect each apple individually and report the count automatically.
02 · THE SOLUTION
Colour and depth, combined.
The system uses synchronized colour and depth images so it can separate each apple, even when they sit close together.
- 01
Capture colour and depth
Synchronized colour and depth images are captured of the fruit on the line.
- 02
Detect each apple
The model identifies every individual apple in the frame.
- 03
Draw each boundary
Each apple is segmented so its boundary is clearly defined.
- 04
Report the count
The number of detected apples is reported live.
03 · KEY CAPABILITIES
What the system does.
Colour and depth imaging
Combines synchronized colour and depth images for a fuller view of the fruit.
Individual apple detection
Finds each apple separately, with a confidence score for every detection.
Boundary segmentation
Outlines the boundary of each apple.
Live counting
Reports the number of detected apples in real time.
04 · OUTCOME
Automatic counting on a production line.
The project delivered a detection system that identifies each apple, draws its boundary and reports the count live.
It shows how colour and depth imaging can take a repetitive visual counting task off people's hands.
FREE AI AUDIT
Have a complex workflow worth solving?
Tell us about the process slowing your team down. We'll help you assess whether a custom AI solution could make a difference.