HALCON allows customers to use Deep Learning technologies in their own applications and even provides free tools to improve efficiency.
Deep Learning opens a new market for vision that was previously closed due to complexity using traditional machine vision.
Applicability
Where is it well suited?
- Defect Detection
- Identification
- Grading
Automating menial tasks
- Picking weeds
- Replacing manual visual inspection
Automating highly skilled tasks
- Grading Alfalfa Hay
- Grading wood stain colors
- Especially applications where traditional MV isn’t capable or too difficult
Image Classification
- Assign an entire image to a trained class
- Individual images labeled via the file structure
- Simple approach
- Can workaround the limitations by tiling images
Object Detection
- Identify and locate objects with a bounding box
- Improve labeling and training
- Allow counting and simple post-processing tasks
- Axis aligned versus object aligned
Semantic Segmentation
- Classify pixels within an image as belonging to a specific class
- Allows significant advances in image post processing
- Allow integration with, for example, vision-guided robotics
Anomaly Detection
- Possibility to perform an inspection using only a relatively low number of “good” images for the training.
- Inference results in the “anomaly” that was detected in the inspected image compared to the trained images.
