Hadoop face detection system

Technologies

• Python
  • PyArrow - Apache Arrow Python
• Hadoop
  • Hadoop streaming
• OpenCV
  • Haar cascade algorithm

Pipeline

Mapper

MapReduce Job Execution Flow

Evaluation

We have used F1-score as metric for evaluation.

The average result on testing dataset was 68,40%

For the purpose of increasing accuracy, we may utilize a larger dataset and a more complex face detection algorithm, such as YOLO.