Optimize preprocessing of image for the googlenet-v2 Image Classification Model with Preprocessing API in OpenVINO™
This tutorial demonstrates how the image could be transform to the data format expected by the model with Preprocessing API. Preprocessing API is an easy-to-use instrument, that enables integration of preprocessing steps into an execution graph and perform it on selected device, which can improve of device utilization. For more information about Preprocessing API, please, see this overview and details. The tutorial uses InceptionResNetV2 model.
The tutorial consists of the following steps:
- Downloading the model
- Setup preprocessing with ModelOptimizer, loading the model and inference with original image
- Setup preprocessing with Preprocessing API, loading the model and inference with original image
- Fitting image to the model input type and inference with prepared image
- Comparing results on one picture
- Comparing performance
If you have not installed all required dependencies, follow the Installation Guide.