Lab 5 Video Classification and Object Detection - meetsriharsha/RTBDA_5543 GitHub Wiki
#Welcome to the RTBDA_5543 wiki!
09/28/2016
##Description:
In this lab exercise, we developed a video classification application that is capable of classifying the image in the given video stream.
We have extracted features out of the video. Apart from that, we have extracted features for each object that could be possibly be there in the given input video. These extracted features is provided to the spark as a training model and is tested against the testing model for Image Classification. We have used Decision Tree algorithm for Image Classification and has produced the results in the form of confusion matrix.
Snapshots:
The above snapshot shows the confusion matrix. Usually confusion matrix is used to describe the performace of a classification model. In the above results, we have the diagonal value to be 1 which is a very accurate model.
In the above snapshot, F-Measure, Precision, Recall has been calculated using the metrics class.
Upon classification, we have found the person Johnny as per the training set.
Here is the second image classification. It is able to detect the person Rachel as herself. But in this case, our model is considering the background of the first image and is detecting the second scene as two different images.
Panopto Video Link: