I am Chris from WebsiteLearners.com. We're a media company where we edit tutorial videos.
As we have a lot of new video editors on the team, we need a quick & easy way to give feedback on the edited videos.
We need an AI program to point out common issues that we see in new video edits.
Video Guidelines:
If any point in the guideline is missed in the new video, we want to add a comment with the corresponding timecode where the issue occurred. Like the image attached below.
The AI program can send the comments to this app via [frame.io](http://frame.io) API.
I’ve shortlisted the top 3 issues here, which we can work on the 1st phase of the project.
Please have a look and let me know the time & cost needed to complete the 1st phase.
Thank you.
______________________
P.S. Regarding the sample data,
We have around 40-50 videos, with feedback manually added by our team members.
* More Details:
Please note that all videos will be in 4K resolution & 60 fps. We need the comment to be posted on the video along with the timecode (where the issue occurs). Once the program is ready, we will have to connect it with the [frame.io](http://frame.io/) API to add the comments.
This project is a multi-dimensional. Let me explain its dimensions.
To solve first two issues (This & Here and Box on verbs)
1) We first need to analyze the voice over, i.e convert voice to text.
This is "speech to text conversion" domain of AI.
2) Secondly, we need to perform "Textual Context Analysis" to understand
context of voice over (whether it is a verb or speaker is pointing something, like "this & here".
3) Then Finally comes "Image Processing", where we will look for respective
(expected) behaviour in the video. For example, if text is a verb then
we will look for box (the green/yellow box). If it is not found in the video
(at that specific spot where we performed steps 'a' and 'b'), we will mark this time
step (in video) as mistake.
We need the comment to be posted on the video along with the timecode (where the issue occurs).
Once the program is ready, we will have to connect it with the frame.io API to add the comments.