The Intelligence Advanced Research Projects Activity has asked industry to submit proposals for the development of an automated tools to detect activity, objects and people in multi-camera streaming videos.
IARPA said in a broad agency announcement posted Sept. 13 on FedBizOpps that it plans to implement Deep Intermodal Video Analytics tools for forensic analysis and real-time alerting applications.
The three-phase DIVA program also seeks a common framework and a software prototype that will work to analyze primitive activities that occur in ground-based videos, complex activities and person/object across multiple camera viewpoints,
IARPA noted the framework should be able to support open cloud architecture.
The agency expects consortia of academic institutions and companies worldwide to participate in the DIVA program and that participants will share their expertise in areas such as machine learning, hierarchical modeling, artificial intelligence, object, person and video activity detection, recognition and three-dimensional reconstruction from video.
IARPA will field questions from interested parties through Oct. 12 and accept proposals until Nov. 7.