Computer Vision , Machine Learning , automate video editing based on idioms and template.

We are investigating the viability of computer vision assisting us with the streamlining of video editing. Each video is currently hand edited along the same template. There is 6 clips and 1 key clip which is approx 1 min 30 s long but has a key " segment of about a minute . I believe this key segment can be identified by identifying key events in the clip . These events will ordinarily be identified by a change of light levels ( camera moving from inside to outside ) , angle of key elements ( the sky starting to fill the frame ) and sound level. We are unsure how viable financially this process would be. If you are a machine learning and computer vision expert we would be interested to send you a few video clips and have you tell us how much it would cost for you to auto detect the particular clip ( from a total of five with a templated order with increasing file number .ie all videos are recorded in order evant a has a lower file number than event b etc etc ) and within that clip identify key events or a combination thereof. IF your technique is successful we would integrate it into an automated editing process for which you would also be able to work on.. Something like Magisto but more specialised and templated . We are excited to hear from you. When we do we will send you some typical video clips and describe in more detail our requirements . Perhaps that quick conversation would be the start of a longer term working relationship. At this point budget is generous but I have concerns about the viability or reliability of automated event detection and auto editing following the idioms of the template.