What is Computer Vision?
Computer VisionDeep LearningDroneSegmentationIndustrial AI
What is Computer Vision?
Computer vision is the branch of AI that extracts meaningful information from camera, drone or satellite images. Today it is mostly done with deep learning models.
Core Tasks
- Classification: what is the whole image? ("defective", "OK")
- Object detection: where are the objects, and how many?
- Segmentation: the outline of each object or each pixel. See Instance Segmentation.
- Vectorisation: turning pixel masks into lines and polygons, e.g. CAD drawings from orthophotos.
Industrial Use
- Agriculture: tree counting and health tracking in drone imagery.
- Mapping: automatic drawing of buildings, roads and boundaries from orthophotos.
- Manufacturing: surface defects, missing parts and label checks on the line.
- Safety: detecting helmets, vests and danger-zone intrusions.
Challenges in the Field
A model that scores well in the lab can drop under different lighting, a different camera or a different region. That is why testing must happen under conditions not seen in training. See Overfitting.
Our services: Industrial AI