The metadata-first video pipeline

Real-time video search with state-of-the-art Machine Learning that leaves video in situ

fedex at home office

Real-time video search

Get the advantages of cloud-based Machine Learning without having to upload high resolution video. GanzCloud analyzes video on your local network so that you can upload only the metadata extracted from video events.

In metadata-only mode, GanzCloud enables each 5 Mbps uplink Internet connection to support real-time video search of 150 cameras. So even large corporate campuses and commercial buildings can get state-of-the-art video surveillance using the bandwidth of a typical cable Internet connection.

Best of local storage & cloud intelligence

Notice that the 2,000 kbps video stream remains on your local network. GanzCloud Box analyzes the video in a 30 kbps chatty conversation with the cloud that learns what’s important to you.

When you search for something, GanzCloud returns search results that reference the earliest and latest timestamps of the original video segment so that the video can be served from your own proxy server. For example, the request:

GET https://camio.com/api/search/?text=fedex+home+office
returns results like:
[
	{
	  "source": "Home Office",
	  "cover_image_id": "ag1zfmNhbWlvbG9nZ2VyclILEg9VcGxvYWRl",
	  "earliest_date": "2017-02-26T20:28:25.000-0000",
	  "latest_date": "2017-02-26T20:29:13.000-0000",
	  "labels": [
	      "fedex",
	      "van",
	      "white"
	  ],
	  "images": [
	      {
	        "content_type": "image/jpeg",
	        "date_created": "2017-02-26T20:28:25.000-0000",
	        "id": "ag1zfmNhbWlvbG9nZ2VycksLEg9VcGxvYWRlZENv"          
	      },
	      ...
	  ]
	},
	...
]
					

The help article “How do I tune my GanzCloud Box VM to match the number of connected video streams” provides guidance on scaling to any number of streams. If you’d like help with implementing your own video proxy server, just contact us.