I have collected about 49.000 film footages, each of which belongs to one of the seven classes depending on the shooting plan:

• Extreme Close Up
• Close Up
• Medium Close Up
• Medium
• Medium Wide
• Wide
• Extreme Wide

You can train your own frame classification model and use it in your own projects related to content analysis, metadata creation, automation of editing processes, etc. I hope this data will be useful for you ^_ . https://www.kaggle.com/datasets/minutcoff/movie-images-by-types-of-shooting-plans

I have saved several notebooks and python files in this repository. Some are useful for further data collection, in others I train the efficientnet_v2_s model and analyze model errors.

During the study, it turned out to achieve an average accuracy of 0.7314. At first glance, the result is not impressive, but I want to note that if you ask a person to solve this problem, they are likely to face great difficulties. Below you can see some of the images in which the model was mistaken during the validation process: