Social media has become a key platform to share news and information related to public events e.g., road accidents and crime scenes. Images related to these events can reveal some interesting facts about the incident. Reconstructing a scene based on these crowdsourced images can help to better investigate an incident. Existing work on scene analysis and reconstruction is based on image processing. These techniques are costly in terms of computation. There exist some service-oriented approaches for scene reconstruction as well. A fundamental assumption in these works has hitherto been that the crowdsourced images are trustworthy. In this project, we propose a more objective approach relying on the modifications within the image to assess their trustworthiness. We only utilize the image metadata to assess the trustworthiness of the image. Hence, the computation and resources required will be less as compared to image processing.
