Audio or Video Content Detection
$5000-25000 USD
Paid on delivery
We are looking for a software component developed, capable of preparing a list of "markers" to be used as reference, and then scan a provided file and identify which parts of the reference markers exists in the scanned file. The results of such a scan may look like:
Scanning [url removed, login to view]
Marker "Die Hard" detected at 02:26:31.170 for 00:03:27.807
Marker "Olympics" detected at 01:48:00.000 for 02:74:36.018
The process can use audio or video processing (only one is needed, if you can provide separate algorithm for each, please specify so in your bid), but it cannot rely on the existance of some hidden sign or inserting one to be detected later ("watermark").
While not required at this point, if you can provide expected processing rate it will be appreciated, for example "will scan 1000 seconds of content each processing second on a single-core Pentium 4 3GHz when using 10000 seconds of reference markers".
Open source may be used, but it must be mentioned in the bid. The developed algorithm Intellectual Property must belong to us after development, and the winning bidder will be requested to assist in applying for patents, where applicable, on our behalf as part of this project (the patent application will be handled by us).
There will likely be significant amount of follow-up work for successful coder.
Project ID: #3900220