The volume of video archives is skyrocketing. To use these archives effectively, there is a growing demand for intelligent softwares to automatically understand the video contents. Although it is trivial for human, empowering computers to recognize video semantic concepts looks very ambitious goal. However, walking on the moon was a dream one day which came true through diligence and persistence of some brave scientists. Nothing is impossible! Fortunately, recent advances in artificial intelligence let learning from big datasets and pave the way to develop softwares understanding videos. Sensifai is already developing this technology!
Broadcast companies usually deal with huge archives of videos and usually require re-use of archival material. Sensifai, can provide high-speed search solution for the broadcast professionals. This innovative search method improves the searching options and increases the access to the large video archives. Each uploaded video to a video-sharing website is associated with a user-generated tag(s). Sensifai chunks the video to its different scenes and then automatically generates the tags for each chunk. The generated tags can be indexed in the video search engine to improve the search quality and user satisfaction.
Video-sharing, broadcasting companies and IPTV providers usually add the advertisement to the videos. These Ad-hoc advertisements in videos are usually distractive and unpersuasive. Sensifai provides an automatic advertisement solution such that it automatically finds the change points of a given video to show the ad in those parts to avoid distracting the audiences and then it engages video audiences with ads related to the content of the video.
Broadcast professionals and surveillance system users usually require re-use of archival material. To browse the archival videos quickly, the intelligent video summarization of Sensifai can generate a well-organized and concise version of a given videos based on user preference specified by text or optionally interaction. This lets the users to quickly overview a video database and spot the interested video segments.In Bing video, a short summary of the video is shown by keeping the cursor on the videos to yield an idea about the video content before watching it. However, these summaries are made without any intuition about the video contents. Sensifai can provide an information-reach summary of important events in a video by understanding their content. This helps the users to skim the videos effectively.
Security professionals usually have to monitor live security videos constantly in public places, e.g. metros and airports, to prediction threats and alert in the case of problems such as abnormal human activity, unattended bags, gun shooting, etc. Sensifai can be used to automatically identify abnormal human behavior, and dangerous objects such as a gun, knife, fire, etc. and predict the threats and provide alerts. Video sharing websites usually have to monitor the user uploaded contents to avoid Spam contents or illegal contents, e.g. child pornography and abuse, violent content, etc. Sensifai provides a concept-aware monitor system of submitted content to prevent Spam videos or illegal videos. Furthermore, brands can also use this technology to protect themselves.
When a user watches a video or listen to a music in a video-sharing or a music-sharing website, the website automatically recommends similar videos or music tracks to the user for watching. The conventional video recommender systems usually rely on the tags of videos provided by their users or the history of user activities. Sensifai can improve the recommender systems by incorporating the semantic concept of videos or music mood and emotion.
We are a technical team that is re-defining how machines understand our world