2022-04-19, 09:31
Name: Arvind N
forum/e-mail: *removed by moderator*
Personal background: I work as a C++ and web technologies developer in a company and have used Kodi extensively for watching many TV shows and movies using the Exodus and Krypton plugins. I have great respect for the Kodi project and am excited at the opportunity to be able to contribute to this project!
Summary: In-video feature to search for characters from video being watched and their background
How will I achieve this: The idea of the project is to allow people who are binge-watching shows to pause and search for the character and their wikipedia information from within the video viewing interface itself. I have personally faced this issue many times where I am watching a show but cannot place where the character is from. So, I pause, take screenshot and then go to Google image search and search for the person. This is a time taking process. instead if we can have an add-on that can do this in background, then this will make the video viewing experience even better. This is similar to the in-video data feature that is offered by Amazon prime Video.
What will the project focus on: The project will be broken into parts:
1. Gather data to be searched
Here, we can start with a crude method where clicking on the add-on will pause the video and prompt user to select which character he wants to search about using snapshotting tool. Then, the selected snap will be sent to the API to perform search operation
2. Perform search operation
Image search can be done using Google Images or any other open source image database online (like IMDB) and output data can be parsed to reveal the celebrity or character name. Once this is received. we can even connect this search and take it one step further by searching wikipedia and other sources for more related information about this character. All this information can be tagged and stored for processing.
3. Display search results in interface
All the information returned from the search operation can be parsed, filtered and display on-screen in the video viewing interface itself on demand. And, since the search is done, we can cache this info as well so that it is available easily and faster when it is requested the next time by the viewer.
An additional layer that can be added to this project is to have an initial search be done using the name of the video content itself. This can scrape the internet for trivia, cast, characters and other basic information of the video without any specific scene context given. This will be also cached for future viewing of the same content by the viewer. We can even provide an option for the viewer to control this caching behavior so that Kodi does not eat into the viewer's disk space.
Benefits: This will give an added layer of sophistication to the video viewing interface and enhance the user experience. Especially for binge watchers like myself, this would be a wonderful tool to keep ourselves informed and engaged with the video viewing experience while also getting our trivia game on! This feature can and will bring us on par and to the same level as Amazon Prime Video.
Goals: The main goal of this project should be to set up an addon that can search the internet and display information to the viewer within the video viewing interface. Now, whether this search is triggered by image search or context search or text search, that is the modularization which we can look into as we move further into the project.
What does it touch in Kodi: This will touch primarily only the front end code that deals with the user interface of the video viewing application. All the web scraping can be managed in backend using web workers in JavaScript and cached into TXT files that can be sourced upon as required.
Requirements: This project would require the knowledge of the Kodi user interface (written in C++) and also web scraping technologies (using Javascript or Python, as required. This bridging can be done from C++ as well if needed!)
Possible mentors: Place to add possible mentors (Team-Kodi will add this).
Workload: 350 hours.
forum/e-mail: *removed by moderator*
Personal background: I work as a C++ and web technologies developer in a company and have used Kodi extensively for watching many TV shows and movies using the Exodus and Krypton plugins. I have great respect for the Kodi project and am excited at the opportunity to be able to contribute to this project!
Summary: In-video feature to search for characters from video being watched and their background
How will I achieve this: The idea of the project is to allow people who are binge-watching shows to pause and search for the character and their wikipedia information from within the video viewing interface itself. I have personally faced this issue many times where I am watching a show but cannot place where the character is from. So, I pause, take screenshot and then go to Google image search and search for the person. This is a time taking process. instead if we can have an add-on that can do this in background, then this will make the video viewing experience even better. This is similar to the in-video data feature that is offered by Amazon prime Video.
What will the project focus on: The project will be broken into parts:
1. Gather data to be searched
Here, we can start with a crude method where clicking on the add-on will pause the video and prompt user to select which character he wants to search about using snapshotting tool. Then, the selected snap will be sent to the API to perform search operation
2. Perform search operation
Image search can be done using Google Images or any other open source image database online (like IMDB) and output data can be parsed to reveal the celebrity or character name. Once this is received. we can even connect this search and take it one step further by searching wikipedia and other sources for more related information about this character. All this information can be tagged and stored for processing.
3. Display search results in interface
All the information returned from the search operation can be parsed, filtered and display on-screen in the video viewing interface itself on demand. And, since the search is done, we can cache this info as well so that it is available easily and faster when it is requested the next time by the viewer.
An additional layer that can be added to this project is to have an initial search be done using the name of the video content itself. This can scrape the internet for trivia, cast, characters and other basic information of the video without any specific scene context given. This will be also cached for future viewing of the same content by the viewer. We can even provide an option for the viewer to control this caching behavior so that Kodi does not eat into the viewer's disk space.
Benefits: This will give an added layer of sophistication to the video viewing interface and enhance the user experience. Especially for binge watchers like myself, this would be a wonderful tool to keep ourselves informed and engaged with the video viewing experience while also getting our trivia game on! This feature can and will bring us on par and to the same level as Amazon Prime Video.
Goals: The main goal of this project should be to set up an addon that can search the internet and display information to the viewer within the video viewing interface. Now, whether this search is triggered by image search or context search or text search, that is the modularization which we can look into as we move further into the project.
What does it touch in Kodi: This will touch primarily only the front end code that deals with the user interface of the video viewing application. All the web scraping can be managed in backend using web workers in JavaScript and cached into TXT files that can be sourced upon as required.
Requirements: This project would require the knowledge of the Kodi user interface (written in C++) and also web scraping technologies (using Javascript or Python, as required. This bridging can be done from C++ as well if needed!)
Possible mentors: Place to add possible mentors (Team-Kodi will add this).
Workload: 350 hours.
