hrithik-jha/image-info

Selecting a thumbnail from a video based on user genre preference.

Thumbnail Personalization

Running

Currently all facets of the program are set for 3 clusters.
If the number of clusters has to be changes, change should be introduced in processing/distancing.py, model/predict.py and the shell scripts in processing.

Steps

1. Run the vid2img.py program to partition video into frames.
2. Execute processing/distancing.py to perform frame clustering and create individual file clusters.
3. Find the predictions using model/predict.py for all files in a cluster. Output can be sanitised.

To-do

Allow more than 3 properties to show and choose the ones which are different for each cluster.