Optimization of Real-Time Object Detection on Intel® Xeon® Scalable Processors

This publication demonstrates the process of optimizing TensorFlow object detection inference workload on an Intel® Xeon® Scalable processor for the following objectives:

  1. Achieve real-time frame rate and low latency
  2. Minimize required computational resources

In this case study, a convolutional neural network described in the “You Only Look Once” (YOLO) project is used for object detection. The original darknet model is converted to a TensorFlow model. First, the network is optimized for inference. Then array programming with NumPy and TensorFlow is implemented for post-processing pipeline. Finally environment variables and other configuration parameters are tuned to maximize the computational performance. With the optimizations, real-time object detection is achieved while using a fraction of the available processing power of the benchmark system.

 (DRAFT) Colfax-Real-Time-Object-Detection.pdf (305 KB) — this file is available only to registered users. Register or Log In.

Demonstration: