Identifying & analysing objects
The Situation
Who doesn’t dream of being able to identify and find objects in image files or videos? Or even photographing objects with their cell phone and then using these like a search term?
This dream has now become reality: Apollo is the first – and only – true object recognition technology in the world. Instead of using a search term, a photo with the object searched is used. Apollo recognises this object or its class.
This opens up the following possible applications:
Apollo is thus the search technology of the future – enabling an all-in-one solution for text, image and voice that is unique in the world.
This dream has now become reality: Apollo is the first – and only – true object recognition technology in the world. Instead of using a search term, a photo with the object searched is used. Apollo recognises this object or its class.
This opens up the following possible applications:
- Recognition of products for making available corresponding information to the user
- Direct image search, dispensing with unreliable keyword searches
- Object recognition in multimedia archives
- Voice and speaker recognition, i.e. assignment of a voice to a certain person
Apollo is thus the search technology of the future – enabling an all-in-one solution for text, image and voice that is unique in the world.
Application field
- Using images instead of key words as the “search term”, i.e. searching in image files for similar images
- Face and person recognition under difficult conditions such as angle, mediocre image quality and embedding in groups of persons
- Recognising objects and object categories for product identification systems
- Optical quality control in the high-end products of changing appearance and position, or evaluation of satellite and aerial images, ...
- Monitoring moving objects with automatic segmentation as pre-processing for biometric systems
In this area also, Apollo works extremely fast. The possibilities this open
- Finding images and scenes in very large petabyte* archives. This takes place object-based and not by means of metadata or soundtracks.
- Processing environment images for intelligent control of motor vehicles, industrial vehicles, household appliances, unmanned aerial vehicles, ...
- Finding and identifying objects on the Internet in near real time for security applications, to pursue legal infringements in the digital area, ...
Apollo can process any physical signals, providing an attractive extension for special applications:
- sensor fusion: visible light, infrared, radar and other frequencies can be automatically combined into a single image, opening up new possibilities in the automobile sector.
Handling
What makes Apollo so special is its ease of use. For the user, no programming at all is needed. Instead, Apollo is trained for the respective application. This, too, takes place without any programming in an extremely fast, simple and very flexible process.
When it comes to feature extraction, Apollo – unlike existing well-known systems – works completely independently, performing this difficult task quickly and efficiently. The stored features can thus themselves be searched in extremely large data volumes within a very short time.
When it comes to feature extraction, Apollo – unlike existing well-known systems – works completely independently, performing this difficult task quickly and efficiently. The stored features can thus themselves be searched in extremely large data volumes within a very short time.
* A petabyte is equal to one thousand trillion (1,000,000,000,000,000) bytes. This corresponds to 1,000 terabytes or 1,000,000 gigabytes. According to an estimate of the ECIN, the WWW had a total data volume of approximately 10 petabytes in 2004.
