EVA Project (1993- 1996).
Keywords: Computer Vision. Traffic Monitoring. Industrial Control. Multiprocessing, JPEG.

The objective of the EVA project was to develop a Computer Vision traffic Monitoring system capable of measuring traffic variables such as speed, vehicle counting, traffic composition, etc in highways.

The greatest challenges of the project were: real-time and outdoor conditions. The first problem was solved through the development of a massive parallel processing system enclosed in a rugerised cabinet that was meant to be installed at the side of the road. A B/W video camera placed on top of a pole by the side of the road provided the necessary video input to the system. Two versions of the system were developed: the original one was based on Transputer networks while the second one used more powerful Texas Instruments C40 DSPs.

Working in outdoor conditions required the development of self-adaptive computer vision algorithms that relied heavily on a priori scene geometry information, real time tracking algorithms and fast segmentation techniques.

The system entered into industrial implementation and more that 400 units were installed around the city of Madrid. The commercial system comprised a full-featured remote station composed of Computer Vision modules, real time database, communications, Image Compression module, non-volatile storage, etc.

The system was presented at the International plead of the State of California in 1992 where it scored second best place and first European participant. The system was also evaluated by companies such as: 3M and Huges.