Computer Vision solves your business problems
Computer vision is one of the most prominent fields of artificial intelligence. It enables machines to extract information from digital images and videos.
If artificial intelligence enables computers to think, computer vision enables them to see, observe and understand.
Adapted from IBM 1
Most industries benefit from Computer Vision
Today's imaging devices, such as smartphones, security systems, satellites and traffic cameras, produce immense quantities of digital visual information. This vast amount of data is accelerated by rapid technological growth, enabling computer vision technology's development.
Many operations within various industries benefit from computer vision solutions. These industries include retail, healthcare, energy and utilities, automotive and transportation, and manufacturing. The market for computer vision solutions continues to grow and is expected to reach USD 130.57 billion by 2030 at a compound annual growth rate of 26.3% over the forecast period. 2
Technological leap behind the development of deep learning
Recently, the spread of computer vision solutions in industrial use has been held back by both data availability and technological requirements.
However, today’s generation of data and advances in computer processors enable the application of deep learning algorithms. These algorithms train iteratively on large high-quality datasets to form the foundation of computer vision solutions.
MLOps: Integration of Computer Vision into a business chain
Integrating machine learning solutions, such as computer vision, into a business chain requires dataset management, model deployment and model monitoring. These tasks are often shareable and repeatable throughout an organisation. MLOps (Machine Learning Operations) facilitates the tasks by providing a set of best practices.
The best practices improve the collaboration and communication between data scientists and operation professionals during the business integration of machine learning.
AFRY X Data platform introduces Computer Vision to your organisation
Within the AFRY X Data platform (AXD), we have built a computer vision solution for different tasks. These tasks currently include image classification, object detection and image segmentation. This computer vision solution can be deployed as a service to any organisation.
When building a specific solution, we use transfer learning to implement a previously trained model and modify it as needed for our application. This way, we can significantly reduce the time and cost required to prepare and tune the model for a client’s specific computer vision solution.
We use Apache Spark within Databricks when training computer vision models for customers with a vast number of images or limits to the individual image size. Additionally, we use Databricks-supported libraries that facilitate parallelised computation, such as Petastorm, Hyperopt and Horovod.
Our solution is built with MLOps in mind. We use Terraform to deploy cloud infrastructure and create data-storage pipelines, set up machine-learning environments, configure workflows to run code, store trained models and host models on a server. The data workflow involves data annotation from the source and preprocessing based on the specific computer vision task. The end-to-end model-development workflow includes model training (using pre-trained models), model staging, comparison of production and staging models, model deployment and model serving.