rocAL Unit Tests#
This application can be used to verify the functionality of the API offered by rocAL.
Build Instructions#
Pre-requisites#
Ubuntu Linux, version
16.04
or laterrocAL library (Part of the MIVisionX toolkit)
ROCm Performance Primitives (RPP)
Python3
Pillow
Install Pillow library using python3 -m pip install Pillow
Build#
mkdir build
cd build
cmake ../
make
Running the application#
./rocAL_unittests
Usage: ./rocAL_unittests reader-type pipeline-type=1(classification)2(detection)3(keypoints) <image-dataset-folder> output_image_name <width> <height> test_case gpu=1/cpu=0 rgb=1/grayscale=0 one_hot_labels=num_of_classes/0 display_all=0(display_last_only)1(display_all)
Output verification#
The bash script testAllScript.sh
can be used to run and dump the outputs for all test cases in rocAL and run the python script to verify the correctness of the generated outputs with the golden outputs.
Input data is available in the following link : MIVisionX-data
export ROCAL_DATA_PATH=<absolute_path_to_MIVIsionX-data>
./testAllScripts.sh <device_type 0/1/2> <color_format 0/1/2>
Device Type
Option 0 - For only HOST backend
Option 1 - For only HIP backend
Option 2 - For both HOST and HIP backend
Color Format
Option 0 - For only Greyscale inputs
Option 1 - For only RGB inputs
Option 2 - For both Greyscale and RGB inputs