Live Video Analyzer

We are looking at developing a small prototype which can extract data from live video feeds and put it in a simple DB (or CSV). The software would take video feed as input and give structured data as the output. The data to be extracted should be configurable (using config/xml, etc.). The data can be like, e.g., detection of number of specific objects in the video, etc, like people, cars, or any other object as defined through the config file. More details below: 1. The solution shall comprise of a Web-based application which will extract data from either a live video stream or any video file supplied to it. 2. The live video stream can be from any camera connected to the Laptop/PC. There will not be any limitation on the following: a. The type of camera used b. The angle at which the camera is placed 3. The extracted data shall be saved to a database (or CSV file) 4. The application shall detect the following objects in the video stream: a. Humans b. Cars The above two object types shall be configurable using an XML file 5. Accuracy: The accuracy of detection of these two types of objects shall be a minimum of 90% 6. Performance: The system should be performing very efficiently and detect and log objects as they are found, while showing the live video feed. 7. The data shall be stored in the following form: a. Date b. Time c. Camera ID d. Object Type (Person / Car) e. Count 8. Each time the object(s) is detected, a row shall be stored in the DB as per point #7 above 9. Deliverables: a. A fully working application for Windows platform b. All supporting third party libraries, etc. required to run the application. c. Complete source code in working condition d. The code will have detailed comments Please provide a detailed technology stack and high level solution architecture (write-up only) while placing your bid. Please also provide a reference to a similar kind of work done by you. While placing your bid, it will be great if you can provide a screencam of a similar solution in working form. Preferred Technology: OpenCV