An Object Detection and its Educational Effect on Aspiring Film Directors using Computer Vision and Machine Learning

Le résumé fourni par la source

The issue that is being addressed is how those who wish to be involved in film production in the future can be more educated regarding what goes into making a movie. An application was created to help these people keep track of movie props and characters and where they are placed throughout a movie. This is performed by using videos or images as input for the application, and the application uses an object detection model from MediaPipe to determine what types of objects are in a scene and where exactly they are located in the scene [4]. Then, bounding boxes are placed around the objects with the classification of the object next to each bounding box, and the result is returned. The results of the experiment indicate that the application is consistent at detecting objects when the object is fully shown on the screen. However, when the application used the cropped images as input as opposed to the full version of the image, the application was not consistently accurate, as it was not able to recognize objects that were partially cut out of the picture. More work will be needed to improve both the object detection capabilities of partially obstructed objects and the overall user interface of the application [5].

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.