Executive Summary : | Preserving law and order and aiding people who suffer due to accidents is one of the major challenges for the government in India, particularly state goverments (such as Uttar Pradesh). Most crimes and accidents that happen at night are harder to address as the general public movement is rarer. During daytime many untoward incidents are averted due to the presence of people. Video surveillance is a fairly well researched area with good solutions that detect and track people and identification of abnormal events possible in well-lit environments. However, these are not so well developed for night or adversely illuminated conditions. There are now IR cameras available that can be used for surveillance. The aim is to obtain good algorithms that can be adapted to work with the night vision cameras. Further, we also aim to transcribe the events into natural language. This will help in automatic messaging of alerts or events to the relevant persons. This could for instance be used for improving the predictive patrolling by the state police ( for instance: UP 100) as it would automatically send alerts regarding untoward activity at night. Once the technology is developed it can be deployed widely to deter criminal activity and also be used to aid people in case of accidents on roads at night by sending the relevant information to highway patrols. This technology would also be relevant to Indian defense organizations for surveillance uses such as patrolling the border or surveillance of sensitive installations. |