Robustness! This is the keyword of our ID recognition technology.

Images obtained from real world conditions are unlike those obtained in labs. Firstly, we have limited control over conditions like lighting, shadows, and the weather. Secondly, input obtained from the real world is never perfect. Characters are smeared, badly printed, slanted, or occluded. An ID recognition system that is expected to work in real world applications must be designed to handle all the above conditions. This is where robustness comes in. By skilfully capitalizing on ISS's expertise in neural network and image processing technologies, we are able to achieve that robustness.

Our ID recognition system was designed and developed to work in adversed real-world conditions. It can work in conditions of rain or shine, day or night, outdoors or indoors. It can also handle "noisy" input, charaters that are smeared, badly printed, slanted, or occluded. Our system can give performances comparable to what others achieve in controlled lab conditions.

Indeed, our technology has been used successfully in real world applications. We have installed 19 ID recognition systems in PSA container gates for verification of contaider IDs. It has achieved a 95% performance at the worksite. We are currently working on a similar system on Quay Cranes. This system is expected to return potential savings of up to $10 million a year.

Our technology can also be applied to other areas. We are currently working on applying the technology to the recognition of vehicle license plate numbers and recognition of serial numbers on computer chips and other devices. Preliminary results of these efforts are very promising.