openvino_python_samples

openvino_python_samples copied to clipboard

Introduction

This repo contains couple python sample applications to teach about Intel(R) Distribution of OpenVINO(TM).

Object Detection Application

openvino_basic_object_detection.py

This applications intends to showcase how a model is being used with OpenVINO(TM) Toolkit.

Aim is to show initial use case of Inference Engine API and Async Mode.

usage: openvino_basic_object_detection.py [-h] [--model-xml MODEL_XML]
                                          [--model-bin MODEL_BIN]
                                          [--target-device TARGET_DEVICE]
                                          [--input-type INPUT_TYPE]
                                          [--input INPUT]
                                          [--detection-threshold DETECTION_THRESHOLD]
                                          [--async]
                                          [--request-number REQUEST_NUMBER]

Basic OpenVINO Example with MobileNet-SSD

optional arguments:
  -h, --help            show this help message and exit
  --model-xml MODEL_XML
                        XML File
  --model-bin MODEL_BIN
                        BIN File
  --target-device TARGET_DEVICE
                        Target Plugin: CPU, GPU, FPGA, MYRIAD, MULTI:CPU,GPU,
                        HETERO:FPGA,CPU
  --input-type INPUT_TYPE
                        Type of Input: image, video, cam
  --input INPUT         Path to Input: WebCam: 0, Video File or Image file
  --detection-threshold DETECTION_THRESHOLD
                        Object Detection Accuracy Threshold
  --async               Run Async Mode
  --request-number REQUEST_NUMBER
                        Number of Requests

mobilenet-ssd Model Download & Conversion

Downloaded model with OpenVINO(TM) Toolkit Model Downloader

python3 /opt/intel/openvino/deployment_tools/tools/model_downloader/downloader.py \
--name mobilenet-ssd \
--output_dir ~/openvino_models/

Converted Model using OpenVINO(TM) Toolkit Model Optimizer

• FP32 Conversion

python3 /opt/intel/openvino/deployment_tools/model_optimizer/mo.py \
--input_model ~/openvino_models/object_detection/common/mobilenet-ssd/caffe/mobilenet-ssd.caffemodel \
--output_dir ~/openvino_models/object_detection/common/mobilenet-ssd/FP32 \
--data_type FP32 \
--scale 127.50223128904757 \
--input_shape [1,3,300,300] \
--mean_values [127.5,127.5,127.5] \
--framework caffe

• FP16 Conversion

python3 /opt/intel/openvino/deployment_tools/model_optimizer/mo.py \
--input_model ~/openvino_models/object_detection/common/mobilenet-ssd/caffe/mobilenet-ssd.caffemodel \
--output_dir ~/openvino_models/object_detection/common/mobilenet-ssd/FP16 \
--data_type FP16 \
--scale 127.50223128904757 \
--input_shape [1,3,300,300] \
--mean_values [127.5,127.5,127.5] \
--framework caffe

Face Detection Application

openvino_basic_object_detection.py

This application intends to showcase how two model from Open Model Zoo (https://github.com/opencv/open_model_zoo) is being used.

Aim is to show initial use case of Inference Engine API Dynamic Batching with using Face Detection and Gender-Age Detection.

usage: openvino_face_detection.py [-h] [--face-model-xml FACE_MODEL_XML]
                                  [--face-model-bin FACE_MODEL_BIN]
                                  [--face-detection-threshold FACE_DETECTION_THRESHOLD]
                                  [--face-target-device FACE_TARGET_DEVICE]
                                  [--ag-model-xml AG_MODEL_XML]
                                  [--ag-model-bin AG_MODEL_BIN]
                                  [--ag-max-batch-size AG_MAX_BATCH_SIZE]
                                  [--ag-dynamic-batch]
                                  [--ag-target-device AG_TARGET_DEVICE]
                                  [--input-type INPUT_TYPE] [--input INPUT]

Basic OpenVINO Example to Face, Age and Gender Detection

optional arguments:
  -h, --help            show this help message and exit
  --face-model-xml FACE_MODEL_XML
                        Face Detection Model XML File
  --face-model-bin FACE_MODEL_BIN
                        Face Detection Model BIN File
  --face-detection-threshold FACE_DETECTION_THRESHOLD
                        Face Detection Accuracy Threshold
  --face-target-device FACE_TARGET_DEVICE
                        Target Plugin: CPU, GPU, FPGA, MYRIAD, MULTI:CPU,GPU,
                        HETERO:FPGA,CPU
  --ag-model-xml AG_MODEL_XML
                        Age-Gender Detection XML File
  --ag-model-bin AG_MODEL_BIN
                        Age-Gender Detection BIN File
  --ag-max-batch-size AG_MAX_BATCH_SIZE
                        Age Gender Detection Max Batch Size
  --ag-dynamic-batch    Age Gender Detection Enable Dynamic Batching
  --ag-target-device AG_TARGET_DEVICE
                        Target Plugin: CPU, GPU, FPGA, MYRIAD, MULTI:CPU,GPU,
                        HETERO:FPGA,CPU
  --input-type INPUT_TYPE
                        Type of Input: image, video, cam
  --input INPUT         Path to Input: WebCam: 0, Video File or Image file

Face Detection Model

Downloaded Face Detection model as below:

python3 /opt/intel/openvino/deployment_tools/tools/model_downloader/downloader.py \
--name face-detection-retail* \
--output_dir ~/openvino_models/

Detailed information about the model:

• https://github.com/opencv/open_model_zoo/blob/master/intel_models/face-detection-retail-0004/description/face-detection-retail-0004.md

Age Gender Detection Model

Downloaded Age Gender Detection Model as below:

python3 /opt/intel/openvino/deployment_tools/tools/model_downloader/downloader.py \
--name age-gender-recognition-retail* \
--output_dir ~/openvino_models/

• https://github.com/opencv/open_model_zoo/blob/master/intel_models/age-gender-recognition-retail-0013/description/age-gender-recognition-retail-0013.md

Object Detection Application with Configurations

openvino_object_detection_configs.py

This application is similar to Object Detection Application above, however some additional configurations options added to see the performance and understand their effect to overall performance.

usage: openvino_object_detection_configs.py [-h] [--model-xml MODEL_XML]
                                            [--model-bin MODEL_BIN]
                                            [--target-device TARGET_DEVICE]
                                            [--input-type INPUT_TYPE]
                                            [--input INPUT]
                                            [--detection-threshold DETECTION_THRESHOLD]
                                            [--async]
                                            [--request-number REQUEST_NUMBER]
                                            [--pc]
                                            [--cpu-num-threads CPU_NUM_THREADS]
                                            [--cpu-bind-thread]
                                            [--cpu-throughput-streams CPU_THROUGHPUT_STREAMS]
                                            [--gpu-throughput-streams GPU_THROUGHPUT_STREAMS]
                                            [--gpu-throttle]

Basic OpenVINO Example to Test Object Detection Model

optional arguments:
  -h, --help            show this help message and exit
  --model-xml MODEL_XML
                        XML File
  --model-bin MODEL_BIN
                        BIN File
  --target-device TARGET_DEVICE
                        Target Plugin: CPU, GPU, FPGA, MYRIAD, MULTI:CPU,GPU,
                        HETERO:FPGA,CPU
  --input-type INPUT_TYPE
                        Type of Input: image, video, cam
  --input INPUT         Path to Input: WebCam: 0, Video File or Image file
  --detection-threshold DETECTION_THRESHOLD
                        Object Detection Accuracy Threshold
  --async               Run Async Mode
  --request-number REQUEST_NUMBER
                        Number of Requests
  --pc                  Enable Performance Counters
  --cpu-num-threads CPU_NUM_THREADS
                        Limit CPU Threads
  --cpu-bind-thread     Bind Threads to CPU
  --cpu-throughput-streams CPU_THROUGHPUT_STREAMS
                        Int Values or CPU_THROUGHPUT_NUMA if not set
                        CPU_THROUGHPUT_AUTO
  --gpu-throughput-streams GPU_THROUGHPUT_STREAMS
                        Int Values if not set GPU_THROUGHPUT_AUTO
  --gpu-throttle        multi-device execution with the CPU+GPU performs best
                        with GPU trottling hint

Object Detection Multi Framework Sample

This sample application will be used to compare inference between multiple frameworks; opencv, openvino