Real-time image classifier using Apache Flink

1. Project Objective:

   • Develop an Apache Flink real-time streaming service for a Machine Learning Application.
   • Conduct image classification on the incoming pixel stream from the source.
   • Record the prediction results and send them to the output destination (sink).

2. Secondary Goal:

   • Compare the performance of Apache Flink with a Python script for the image classification task.

3. Experimental Approach:

   • Conduct multiple trials (5 in total) to ensure consistency in results.
   • Calculate the average result from the trials.

4. Conclusion:

   • Findings indicate that the execution time using Apache Flink is faster.
   • Highlight additional advantages, such as parallelization, achieved through Apache Flink.

Results