Computer Vision is about machine analysis of digital images or video. It is particularly good at tasks which most people find boring, tedious and/or time demanding. It can be:
- Sorting and classification of images for the purpose of making them searchable.
- In videos it may to keep an eye on particular events such as the appearance of objects or specific situations.
- Decoding and digitization of photographed text and numbers.
Make the machine keep an eye on things
If you have one or several cameras that routinely photograph an area a Computer Vision application can analyze the material at regular intervals to detect what you finde interesting:
- If you are the organizer of an event and you need to make sure that specific areas are not over-crowded. Are there signs that people are starting to stand too close to each other?
- In the transportation of goods: Counting how many object of certain types are passing a threshold line within a time frame.
- If you are the sponsor of an event that is shown on television you may want to know visible your logo has been during the broadcast?
Digitization of text in photos
To find text and number in photos and transform them in to digital information is a complete topic by itself. Computer Vision techniques play a major role in this field. But it is also a branch within a much wider field that includes topics such as: Language, text analysis, typography, document format, and more.
From a photo of a physical paper document (receipts, A4 paper, book pages, ...) to an improved graphical version (deskewed, white/black contrast, noise removal), OCR (optical character recognition), spelling checking, word composition through to saving as a final PDF, Word/OTF or HTML document.
The development process
Do you have an idea for how Computer Vision can be applied? Then please Clondyke Software for a talk about what is possible. Previous project experience seems to indicate that it is often a good idea to start out with a simple prototype that demonstrates the main concept. Keeping it simple in the initial phase will help keep the development costs at a low level.
In many cases the development of a Computer Vision application can be described a learning process where you are the teacher and the machine is learner. The teaching and learning takes place by you feed the software with a large number of visual samples. The samples are examples of what is inside or outside those particular categories that you find interesting. Gradually as you feed the machine with your knowledge the machine algorithms will (sort of) learn how to distinguish categories in the visual material in ways that so far only humans have been able to.
The fact that you take part in the process yourself will inevitably require some focus and time from you or a colleague. However, in a development process with Clondyke Software have state of the art tools at your disposal to ease the part that many is considered the heavy part of the process.
Here you can see on the Clondyke Software tools in action: