Lab 6 - PoojaShekhar/CS5543-Real-Time-Big-Data-Analytics--Lab-assignments GitHub Wiki

Development of Simple Kafka -Storm Application

  • Implement a client application for your video data, where the features, key frame and metadata is extracted and sent over to Storm using Kafka.

  • The Storm performs different analytics such as number of key frame, number of sift features for particular video on the data received from the Kafka Spout. Make sure to have multiple bolts to perform different analytics.

I implemented this video application without storm in the architecture.

Video --> Consumed by Kafka Producer(Extracts frames of the video-->Extracted sift features and metadata sent to mlab --> Kafka Consumer pulls the data from mlab and decodes/reconstructs the video

At Kafka Producer metadata of the video was generated and posted to mlab.

mlab

The original image had 346 frames and took 15171 milli seconds and our new video had 54 frames and took 525 milli seconds