Semantic Segmentation Application

Table of Contents

• Introduction
• Supported plaforms
• Data flow
• Steps to run the application on J721e EVM
• Steps to run the application on PC Linux x86_64
• Sample Output

Introduction

This example shows how to write an OpenVx application which runs pre-processing, post-processing nodes on C66x DSP and runs TIDL node on C7x DSP for Semantic Segmentation. The demo is validated on tiadsegNet_v2 network. The trained network is imported using TIDL import tool and raw NV12 input files of resolution 1024x512 is read from SD card or USB 3.0 typec-pen-drive or PCIe memory. The input files are scaled using VPAC multi-scaler MSC to produce a DL resolution of 768x384 as required by the network. As a pre-processing step, the scaled NV12 resolution is converted to RGB format and padded as required on C66x_1 DSP. The pre-processed input is given to TIDL running on C7x DSP for inference. As a post-processing step, the TIDL output is read and each pixel is blended with a color map on the provided image on C66x_2 DSP. The final output is blended with system performance statistics and displayed on a 1920x1080 screen.

Supported plaforms

Platform	Linux x86_64	Linux+RTOS mode	QNX+RTOS mode	SoC
Support	YES	YES	YES	J721e

Data flow

Steps to run the application on J721e EVM

1. Build the application and related libraries as mentioned in Build Instructions for Linux+TI-RTOS mode
2. A sample "app_seg.cfg" for Semantic Segmentation is provided under "/opt/vision_apps/" on the rootfs partition.
3. Create a folder and keep all the .yuv files used for detection, there are a few images under "/opt/vision_apps/test_data/psdkra/tidl_demo_images".
4. In case a different network is chosen, update the dl_size accordingly for network input resolution.
5. Run the app as shown below

   ./run_app_tidl_seg.sh

Steps to run the application on PC Linux x86_64

1. Build the application and related libraries as mentioned in Build Instructions for PC emulation mode
2. A sample "app_seg.cfg" for Semantic Segmentation is provided under "${PSDKR_PATH}/vision_apps/apps/dl_demos/app_tidl_seg/config" folder.
3. Create a folder and keep all the .yuv files used for detection, there are a few images under "/opt/vision_apps/test_data/psdkra/tidl_demo_images".
4. In case a different network is chosen, update the dl_size accordingly for network input resolution.
5. Create an output directory

   mkdir app_tidl_seg_out

6. Run the app as shown below

   ./vx_app_tidl_seg --cfg ${PSDKR_PATH}/vision_apps/apps/dl_demos/app_tidl_seg/config/app_seg.cfg

7. The output will be written in "app_tidl_seg_out" folder in .yuv (NV12) format.

Sample Output

Shown below is a example input and its corresponding output