Open Source Re

NYU-NYC StreetScene dataset.

The NYU-NYC StreetScene dataset is a high-resolution, high-quality video dataset with accurate bounding box annotations. The dataset consists of stereo videos captured using a ZED-mini camera on the streets of NYC. Portions of each video were manually annotated with bounding boxes across 15 classes of objects of interest. Objects of interest were selected based on relevance to pedestrian navigation and obstacle avoidance. The videos were all captured in 2.2K resolution (4416x1242) at 15 fps and the bitrate used to encode the videos was 50Mbps, which ensures that compression artifacts are minimal.

VI App Arcade

The BVI App Arcade is a repository of apps developed for those with visual impairments and is broken down into 11 categories, ranging from entertainment to navigation to socializing. The repository emulates a video game arcade; all of the apps are thematically grouped into sections and then each app is broken down into specific features, almost as if the pinball machines were grouped together with other pinball machines, but in this case we are talking about cool mobile applications and what types of neat doodads each app possesses. Every app in the repository has an overview to cover critical features and also a 9 question evaluation rubric or a sort of report card to score each app in the database. The score is based on a 36 point scale and predicated on five key metrics: reputability, privacy, data sharing capabilities, effectiveness, and accessibility. In the BVI App Arcade, we hope you are able to search apps according to your own specific needs, compare and contrast the different apps within each category based on the evaluation metrics, and, most importantly, have fun while searching and learning!