This tutorial [5] requires creation of extensible platform setup as described in TE0950, Vitis 2023.2.1, Extensible Platform [1].

This tutorial [5] requires installation of AMD DPUCV2DX8G for Vitis 2023.2.1 as described in TE0950, Vitis 2023.2.1, Integration of DPUCV2DX8G [3].

The target system is TE0950-02 (with Versal xcve2302 ES1 device) or TE0950-03 (with Versal xcve2302 device).

The TE0950 extensible bring up script are using Linux make utility.
The board configuration files are identical with [6], [7] for the Versal device.
The Artix device is using the bring up scripts and board configuration files [6], [8].
Description of the AMD DPUCV2DX8G can be find in AMD product guide PG425 [9].

Requirements

Download package TE0950_2023_2_1_dpucv2dx8g_AI_3_5.zip accompanying this tutorial from:

https://shop.trenz-electronic.de/trenzdownloads/Trenz_Electronic/Development_Boards/TE0950/Reference_Design/2023.2/VitisAI/TE0950_2023_2_1_dpucv2dx8g_AI_3_5.zip

Install packages

Download the Vitis-AI 3.5 repository.

In browser, open page:

https://github.com/Xilinx/Vitis-AI/tree/master

Click on green Code button and download Vitis-AI-3.master.zip file.
Unzip to

~/work/Vitis-AI-3.5

The directory contains the Vitis-AI 3.5 framework.

Starting point for exploration of these Vitis AI 3.5 examples is this Xilinx www page.

https://xilinx.github.io/Vitis-AI/3.0/html/index.html

Install Docker

In Ubuntu 20.04, install docker

sudo apt install docker.io

Test the docker installation:

sudo docker run hello-world

Install Docker Images for Vitis AI 3.5

Install precompiled docker images

sudo docker pull xilinx/vitis-ai-tensorflow-cpu:latest

sudo docker pull xilinx/vitis-ai-tensorflow2-cpu:latest

sudo docker pull xilinx/vitis-ai-pytorch-cpu:latest

Vitis AI 3.5 Model Downloads

Python Downloader

Vitis AI 3.5 models can be downloaded with python tool downloader.py

cd ~/work/Vitis-AI-3.5/model_zoo

Example:

Select pytorch model framework

input: pt
Select model (vehicle_type_resnet18_pt-vek280-r3.5.0)

input num: 1

chose model (all)

Input num: 0

devel@ubuntu:~/work/Vitis-AI-3.5/model_zoo$ python3 downloader.py
Tip:
you need to input framework and model name, use space divide such as tf vgg16
tf:tensorflow1.x  tf2:tensorflow2.x  cf:caffe  dk:darknet  pt:pytorch  all: list all model
input:pt
chose model

0 : all
1 : pt_vehicle-type-classification_3.5
2 : pt_inceptionv3_0.3_3.5
3 : pt_squeezenet_3.5
4 : pt_face-mask-detection_3.5
5 : pt_OFA-yolo_3.5
6 : pt_OFA-yolo_0.5_3.5
7 : pt_bert-tiny_3.5
8 : pt_resnet50_0.3_3.5
9 : pt_HRNet_3.5
10 : pt_yolox-nano_3.5
11 : pt_pointpillars_3.5
12 : pt_OFA-resnet50_0.88_3.5
13 : pt_resnet50_0.7_3.5
14 : pt_inceptionv3_3.5
15 : pt_OFA-resnet50_0.74_3.5
16 : pt_fadnet_0.65_3.5
17 : pt_fadnetv2_3.5
18 : pt_OFA-resnet50_0.45_3.5
19 : pt_vehicle-make-classification_3.5
20 : pt_bert-large_3.5
21 : pt_vehicle-color-classification_3.5
22 : pt_OFA-resnet50_0.60_3.5
23 : pt_xilinxSR_3.5
24 : pt_resnet50_0.6_3.5
25 : pt_OFA-depthwise-resnet50_3.5
26 : pt_bert-base_3.5
27 : pt_resnet50_0.4_3.5
28 : pt_3D-UNET_3.5
29 : pt_SESR-S_3.5
30 : pt_inceptionv3_0.5_3.5
31 : pt_psmnet_3.5
32 : pt_fadnet_3.5
33 : pt_inceptionv3_0.4_3.5
34 : pt_OFA-rcan_3.5
35 : pt_resnet50_0.5_3.5
36 : pt_OFA-resnet50_3.5
37 : pt_fadnetv2_0.51_3.5
38 : pt_OFA-yolo_0.3_3.5
39 : pt_resnet50_3.5
40 : pt_movenet_3.5
41 : pt_inceptionv3_0.6_3.5
input num: 1
chose model type
0: all
1 : GPU
2 : vek280
3 : v70
input num:0
pt_vehicle-type-classification_3.5.zip 
                                              100.0%|100%
vehicle_type_resnet18_pt-vek280-r3.5.0.tar.gz
                                              100.0%|100%
vehicle_type_resnet18_pt-v70-DPUCV2DX8G-r3.5.0.tar.gz
                                              100.0%|100%
done
devel@ubuntu:~/work/Vitis-AI-3.5/model_zoo$

Delete model packages for v70-DPUCV2DX8G:

./vehicle_type_resnet18_pt-v70-DPUCV2DX8G-r3.5.0.tar.gz

Models for AMD DPUCVDX8H configuration for the TE0950 board have to be compiled from data present in zip archive:

./pt_vehicle-type-classification_3.5.zip

unzip this archive to directory

./pt_vehicle-type-classification_3.5

Arch.json File

An arch.json is produced by Vivado in the process of compilation of the AMD DPUCV2DX8G.

In case of TE0821 system with ID=3, [4] the arch.json file describing the AMD DPUCV2DX8G has been created in:

/home/devel/work/dpucv2dx8g-trd/vitis_prj/hw/binary_container_1/
link/vivado/vpl/prj/prj.runs/prj/prj.gen/sources_1/bd/vu_sys/ip/
vu_sys_DPUCV2DX8G_1_0/arch.json

Copy the arch.json file for the AMD DPUCV2DX8G configuration for the TE0950 to

~/work/Vitis-AI-3.0/model_zoo/arch.json

Compile Pytorch Models

Copy fie compile_model.sh from the repository associated to this application note to the directory

~/work/Vitis-AI-3.5/model_zoo/compile_model.sh

Change mode to executable for file compile_model.sh

Chmod +x ~/work/Vitis-AI-3.5/model_zoo/compile_model.sh

Compile Pytorch Model for AMD DPUCV2DX8G

Change directory to

~/work/Vitis-AI-3.5

Start pytorch docker

sudo ./docker_run.sh xilinx/vitis-ai-pytorch-cpu:latest

In pytorch docker WorkFlow, change directory to model_zoo

cd model_zoo

Start pytorch compiler ftramework

conda activate vitis-ai-pytorch

Parameters for model compilation script compile_model.sh :

pt pytorch model
vehicle_type_resnet18_pt Model name
pt_vehicle-type-classification_3.5 Directory
ResNet_0_int.xmodel Quantized model

Pytorch model is indicated by pt

Model name can be found in the readme file in classification directory:

/home/devel/work/Vitis-AI-3.5/examples/vai_library/samples/vehicleclassification/readme

Part of vehicleclassification/readme file content

Valid model name:
    vehicle_make_resnet18_pt
    vehicle_type_resnet18_pt

Directory is name of the unzipped directory associated to the model

Quantized model is name of file *.xmodel located in directory associated to the model in subdirectory quantized

/work/Vitis-AI-3.5/model_zoo/pt_vehicle-type-classification_3.5/
quantized/ResNet_0_int.xmodel

The file name *.xmodel can be different for different models but the location is in the subdirectory quantized and the extension is .xmodel

Command for compilation of this pytorch model example:

./compile_model.sh pt vehicle_type_resnet18_pt 
pt_vehicle-type-classification_3.5 ResNet_0_int.xmodel

Compilation listing:

vitis-ai-user@ubuntu:/workspace/model_zoo$ conda activate vitis-ai-pytorch
(vitis-ai-pytorch) vitis-ai-user@ubuntu:/workspace/model_zoo$ 
./compile_model.sh pt vehicle_type_resnet18_pt 
pt_vehicle-type-classification_3.5 ResNet_0_int.xmodel
**************************************************
* VITIS_AI Compilation - Xilinx Inc.
**************************************************
[UNILOG][INFO] Compile mode: dpu
[UNILOG][INFO] Debug mode: null
[UNILOG][INFO] Target architecture: DPUCV2DX8G_ISA1_C20B1
[UNILOG][INFO] Graph name: ResNet_0, with op num: 171
[UNILOG][INFO] Begin to compile...
[UNILOG][INFO] Total device subgraph number 3, DPU subgraph number 1
[UNILOG][INFO] Compile done.
[UNILOG][INFO] The meta json is saved to "/workspace/model_zoo/./compiled_output_pt/vehicle_type_resnet18_pt/meta.json"
[UNILOG][INFO] The compiled xmodel is saved to "/workspace/model_zoo/./compiled_output_pt/vehicle_type_resnet18_pt/vehicle_type_resnet18_pt.xmodel"
[UNILOG][INFO] The compiled xmodel's md5sum is 6a4a4c3c1bbc88da22c9415b7f87c7de, and has been saved to "/workspace/model_zoo/./compiled_output_pt/vehicle_type_resnet18_pt/md5sum.txt"
(vitis-ai-pytorch) vitis-ai-user@ubuntu:/workspace/model_zoo$

Classification Model: pt_vehicle-type-classification_3.5 is compiled for AMD DPUCV2DX8G. It contains these 3 files:

~/work/Vitis-AI-3.5/model_zoo/compiled_output_pt/
vehicle_type_resnet18_pt/md5sum.txt
~/work/Vitis-AI-3.5/model_zoo/compiled_output_pt/
vehicle_type_resnet18_pt/meta.json
~/work/Vitis-AI-3.5/model_zoo/compiled_output_pt/
vehicle_type_resnet18_pt/vehicle_type_resnet18_pt.xmodel

One additional file is needed in the compiled model. This file is present in archive with precompiled model for the AMD DPUCV2DX8G.

Open archive

vehicle_type_resnet18_pt-vek280-r3.5.0.tar.gz

copy file

/vehicle_type_resnet18_pt/vehicle_type_resnet18_pt.prototxt

to file

./compiled_output_pt/vehicle_type_resnet18_pt/vehicle_type_resnet18_pt.prototxt

Compile Other Pytorch Models for AMD DPUCV2DX8G

Perform similar compilation steps for all 10 models selected for the demonstration in this application note:

Vehicleclassification of type of vehicle Model (already compiled as an commented example):

1 : pt_vehicle-type-classification_3.5

Vehicleclassification of make of vehicle Model

19 : pt_vehicle-make-classification_3.5

Find related model names in sample application readme file:

~/work/Vitis-AI-3.5/examples/vai_library/samples/
vehicleclassification/readme

Classification of color of vehicle Model:

21 : pt_vehicle-color-classification_3.5

Classification Model:

39 : pt_resnet50_3.5
8  : pt_resnet50_0.3_3.5
27 : pt_resnet50_0.4_3.5
35 : pt_resnet50_0.5_3.5
24 : pt_resnet50_0.6_3.5
13 : pt_resnet50_0.7_3.5

Find related model names in sample application readme file:

~/work/Vitis-AI-3.5/examples/vai_library/samples/classification/readme

Face detection Model:

4: pt_face-mask-detection_3.5

Find the expected model names in sample application readme file:

~/work/Vitis-AI-3.5/examples/vai_library/samples/yolov4/readme

Results of compilations will be in folder

./compiled_output_pt/*

It will be used to target Vitis AI 3.5 demos on TE0950 with AMD DPUCV2DX8G.

Copy this archive into directory indicating configuration for the AMD DPUCV2DX8G like:

~/work/DPUCV2DX8G/compiled_output_pt.zip

Delete content of the directory

~/work/Vitis-AI-3.5/model_zoo/compiled_output_pt/*

Sample Vitis AI 3.5 applications

Demos from the Vitis AI 3.5 library can be compiled in Vitis 2023.2.1 for Petalinux 2023.2 run-time on TE0950 board.

SW examples from the Vitis AI 3.5 package to be moved to TE0950 board are:

~/work/Vitis-AI-3.5/examples/vai_library/samples/vehicleclassification
~/work/Vitis-AI-3.5/examples/vai_library/samples/classification
~/work/Vitis-AI-3.5/examples/vai_library/samples/yolov4

Compress them to:

~/work/vehicleclassification.zip
~/work/classification.zip
~/work/yolov4.zip

Archives to be moved to TE0950 are:

~/work/DPUCV2DX8G/compiled_output_pt.zip
~/work/vehicleclassification.zip
~/work/classification.zip
~/work/yolov4.zip

Compile Tensorflow Models

Vitis AI 3.5 TensorFlow models can be downloaded with python tool downloader.py

cd ~/work/Vitis-AI-3.5/model_zoo

List TensorFlow models

devel@ubuntu:~/work/Vitis-AI-3.5/model_zoo$ python3 downloader.py
Tip:you need to input framework and model name, use space divide such as tf vgg16
tf:tensorflow1.x  tf2:tensorflow2.x  cf:caffe  dk:darknet  pt:pytorch  all: list all model
input:tf

chose model
0 : all
1 : tf_RefineDet-Medical_0.85_3.5
2 : tf_RefineDet-Medical_0.88_3.5
3 : tf_efficientnet-edgetpu-S_3.5
4 : tf_resnetv2-152_3.5
5 : tf_mlperf_resnet50v1.5_3.5
6 : tf_inceptionv3_0.4_3.5
7 : tf_inceptionv3_0.2_3.5
8 : tf_rcan_0.98_3.5
9 : tf_mobilenetv2-1.4_3.5
10 : tf_inceptionv4_0.4_3.5
11 : tf_resnetv1-152_0.51_3.5
12 : tf_inceptionv4_0.2_3.5
13 : tf_resnetv1-101_3.5
14 : tf_superpoint_3.5
15 : tf_mobilenetv2-1.0_3.5
16 : tf_efficientdet-d2_3.5
17 : tf_vgg16_0.5_3.5
18 : tf_inceptionv1_0.16_3.5
19 : tf_mobilenetv1-1.0_0.12_3.5
20 : tf_yolov4-416_3.5
21 : tf_resnetv1-101_0.57_3.5
22 : tf_resnetv1-101_0.35_3.5
23 : tf_vgg16_3.5
24 : tf_vgg19_0.24_3.5
25 : tf_bert-base_3.5
26 : tf_vgg19_0.39_3.
27 : tf_vgg16_0.43_3.5
28 : tf_resnetv1-50_0.65_3.5
29 : tf_efficientnet-edgetpu-L_3.5
30 : tf_RefineDet-Medical_0.75_3.5
31 : tf_resnetv2-101_3.5
32 : tf_mobilenetv1-1.0_3.5
33 : tf_mobilenetv1-0.25_3.5
34 : tf_inceptionv3_3.5
35 : tf_HFNet_3.5
36 : tf_resnetv2-50_3.5
37 : tf_RefineDet-Medical_3.5
38 : tf_resnetv1-152_0.6_3.5
39 : tf_resnetv1-152_3.5
40 : tf_inceptionv1_0.09_3.5
41 : tf_mobilenetv1-1.0_0.11_3.5
42 : tf_ssdmobilenetv1_3.5
43 : tf_ViT_3.5
44 : tf_RefineDet-Medical_0.5_3.5
45 : tf_ssdmobilenetv2_3.5
46 : tf_resnetv1-50_0.38_3.5
47 : tf_mlperf_ssdresnet34_3.5
48 : tf_efficientnet-edgetpu-M_3.5
49 : tf_inceptionv4_3.5
50 : tf_yolov4-512_3.5
51 : tf_yolov3_3.5
52 : tf_resnetv1-50_3.5
53 : tf_vgg19_3.5
54 : tf_inceptionv1_3.5
input num:

Process of compilation models is similar to pytorch model. Change directory to:

~/work/Vitis-AI-3.5

Start tensorflow docker

sudo ./docker_run.sh xilinx/vitis-ai-tensorflow-cpu:latest

In tensorflow docker WorkFlow, change directory to model_zoo

cd model_zoo

Start tensorflow compiler ftramework

conda activate vitis-ai-tensorflow

There are only 3 parameters for model compilation script compile model.sh :

tf tensorflow model
Model name
Directory

The tensorflow model is indicated by tf

Model name can be found in the readme file in the sample application.

Directory is name of the unzipped directory associated to the model.

Compile Tensorflow2 Models

Vitis AI 3.5 TensorFlow2 models can be downloaded with python tool downloader.py

cd ~/work/Vitis-AI-3.5/model_zoo

List TensorFlow 2 models

devel@ubuntu:~/work/Vitis-AI-3.5/model_zoo$ python3 downloader.py
Tip: you need to input framework and model name, use space divide such as tf vgg16
tf:tensorflow1.x  tf2:tensorflow2.x  cf:caffe  dk:darknet  pt:pytorch  all: list all model
input:tf2
chose model
0 : all
1 : tf2_2D-UNET_3.5
2 : tf2_resnet50_3.5
3 : tf2_efficientnet-b0_3.5
4 : tf2_efficientnet-lite_3.5
5 : tf2_inceptionv3_3.5
6 : tf2_mobilenetv3_3.5
7 : tf2_mobilenetv1_3.5
8 : tf2_yolov3_3.5
input num:

Process of compilation models is similar to pytorch model. Change directory to:

~/work/Vitis-AI-3.5

Start tensorflow2 docker

sudo ./docker_run.sh xilinx/vitis-ai-tensorflow2-cpu:latest

In tensorflow2 docker WorkFlow, change directory to model_zoo

cd model_zoo

Start vitis-ai-tensorflow2 compiler framework

conda activate vitis-ai-tensorflow2

There are only 3 parameters for model compilation script compile model.sh :

tf2 tensorflow2 model
Model name
Directory

Tensorflow2 model is indicated by tf2

Model name can be found in the readme file in the sample application.

Directory is name of the unzipped directory associated to the model.

Test Sample Vitis AI 3.5 Applications

Sample application projects are part of the Vitis AI 3.5 package. The download and compilation process will be demonstrated only on a small subset of sample applications for the TE0950 board.

Install Archives on Systems with Petalinux 2023.2

Use WinSCP utility for copy of Vitis AI 3.5 examples:
From PC

~/work/vehicleclassification.zip
~/work/classification.zip
~/work/yolov4.zip

To te0950 board

~/vehicleclassification.zip
~/classification.zip
~/yolov4.zip

Use WinSCP for copy of compiled Vitis 3.5 model archives from PC:

~/work/DPUCV2DX8G/compiled_output_pt.zip

To te0950 board

~/DPUCV2DX8G/compiled_output_pt.zip

On TE0950 board, unzip Vitis AI 3.5 sample applications from

~/vehicleclassification.zip
~/classification.zip
~/yolov4.zip

To directories:

~/vehicleclassification
~/classification
~/yolov4

Copy directory:

~/DPUCV2DX8G/compiled_output_pt/face_mask_detection_pt

To directory:

~/yolov4/face_mask_detection_pt

Copy directories:

~/DPUCV2DX8G/compiled_output_pt/vehicle_make_resnet18_pt
~/DPUCV2DX8G/compiled_output_pt/vehicle_type_resnet18_pt

To directories:

~/vehicleclassification/vehicle_make_resnet18_pt
~/vehicleclassification/vehicle_type_resnet18_pt

Copy directories:

~/DPUCV2DX8G/compiled_output_pt/chen_color_resnet18_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pruned_0_3_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pruned_0_4_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pruned_0_5_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pruned_0_6_pt
~/DPUCV2DX8G/compiled_output_pt/resnet50_pruned_0_7_pt

To directories:

~/classification/chen_color_resnet18_pt
~/classification/resnet50_pt
~/classification/resnet50_pruned_0_3_pt
~/classification/resnet50_pruned_0_4_pt
~/classification/resnet50_pruned_0_5_pt
~/classification/resnet50_pruned_0_6_pt
~/classification/resnet50_pruned_0_7_pt

Compile Vitis AI 3.5 Applications on TE0950 Board

On target board, compile all sample applications:

sh-5.1# cd ~/yolov4
sh-5.1# chmod +x build.sh
sh-5.1# ./build.sh
sh-5.1# cd ~/vehicleclassification
sh-5.1# chmod +x build.sh
sh-5.1# ./build.sh
sh-5.1# cd ~/classification
chmod +x build.sh
sh-5.1# ./build.sh

Test Vitis AI 3.5 Sample Applications on TE0950

Test Vitis AI 3.5 applications.

Commands used for execution of applications are described in sample application readme files:

~/yolov4/readme
~/vehicleclassification/readme
~/classification/readme

Export Path to the DPU

In TE0950 terminal use export command:

On module-based systems supporting modules TE0821 [24] or modules TE0820 [25], use export command:

sh-5.1# export XLNX_VART_FIRMWARE=/run/media/mmcblk1p1/dpu.xclbin

Change directory to

~/yolov4

Test face mask detection with input from file sample_face_mask.jpg

sh-5.1# ./test_jpeg_yolov4 face_mask_detection_pt sample_face_mask.jpg

Output: face mask detection coordinates on terminal.

Test performance of face mask detection with input from files listed in test_performance_face_mask.list file.

sh-5.1# ./test_performance_yolov4 face_mask_detection_pt  test_performance_face_mask.list -s 60 -t 3

Output: data about performance (in FPS) of face mask detection on terminal.

-s 60 defines length of benchmark in seconds.

-t 3 defines use of 3 SW threads (with one AMD DPU)

Test face mask detection with USB www camera input

sh-5.1# ./test_video_yolov4 face_mask_detection_pt 0 -t 1

Face mask detection Vitis AI 3.5 application with camera input, AMD DPUCV2DX8G.

Output: Video output with face mask coordinates in video and performance (in FPS) displayed as text in video.

0 defines USB www camera device number

-t 1 defines use of 1 SW thread

Run test_free_run app. and face_mask app. in parallel

The SD card created finally in [3] contains combination of test_free_run IPs and DPUCV2DX8G unit. While the face_mask detection application is running, go to second TE0950 terminal, and run the test_free_run application by command:

sh-5.1# cd /run/media/mmcblk1p1/
sh-5.1# ./test_free_run dpu.xclbin

Both applications can run in parallel.

The face_mask detection Vitis AI 3.5 application with USB camera input can be stopped by pointing mouse to the X11 video output window and typing Esc key on the keyboard.

Other Vitis AI 3.5 Applications on TE0950 with AMD DPUCV2DX8G

Vitis AI 3.5 sample test applications located in the directory ~/vehicleclassification have analogical format for calling and parameters.

Two different AI models can be used to classify make or type of the vehicle.

Vitis AI 3.5 sample test applications located in the directory ~/classification have analogic format.

Several different AI models can be used to:

Classify color of vehicle,
Classify objects with resnet50_pt model
Classify objects with resnet50_pt model, pruned by factor of 0.3, 0.4, 0.5, 0.6 or 0.7. Models have gradually reduced complexity and potentially reduced classification quality and increased performance.

Measured Vitis AI 3.5 Performance on TE0950 with AMD DPUCV2DX8G

Mesured performance indicators are summarized for TE0950 with AMD DPUCV2DX8G:

Performance in frames per second [FPS],
Power consumption in [W],
End to end performance in (int8) Tera operations per second [Tops].

TE00950 board with AMD DPUCV2DX8G

Vitis AI 3.5 examples	Performance with input from camera e2e [FPS]	Power with input from camera e2e [W]	Performance with input from file e2e [FPS]	Power with input from file e2e [W]	TeraOps with input from file e2e [Tops]
Face detection Model: pt_face-mask-detection_3.5	20.0	13.0	210	13.1	0.14
Vehicle make Model: pt_vehicle-make-classification_3.5	20.0	13.0	730	15.0	2.65
Vehicle type Model: pt_vehicle-type-classification_3.5	20.0	13.0	730	15.0	2.65
Vehicle color Model: pt_vehicle-color-classification_3.5	20.0	13.0	730	15.0	2.65
General classification Model: pt_resnet50_3.5	20.0	13.1	272	14.8	2.23
General classification Model: pt_resnet50_0.3_3.5	20.0	13.1	295	14.6	1.71
General classification Model: pt_resnet50_0.4_3.5	20.0	13.1	317	14.5	1.55
General classification Model: pt_resnet50_0.5_3.5	20.0	13.0	333	14.3	1.36
General classification Model: pt_resnet50_0.6_3.5	20.0	13.0	365	14.2	1.20
General classification Model: pt_resnet50_0.7_3.5	20.0	13.0	428	14.1	1.07

Measurement conditions:

TE0950 with Versal xcve2302 device, 8GB DDR4
IPs in PL fabric: AMD DPUCV2DX8G with interface to AI-ML engines, mem_read, increment, mem_write for test_free_runn application.
USB WWW camera ETERNICO with sensor JX_F23, 1920x1080, max 20 FPS
Power supply 12V/5A
Power measured at the 230V power plug

References

TE0950, Vitis 2023.2.1, Extensible Platform
TE0950, Vitis 2023.2.1, Integration of DPUCZDX8G
TE0950, Vitis 2023.2.1, Integration of DPUCV2DX8G
TE0950, Vitis 2023.2.1, AI 3.0 Models for DPUCZDX8G
TE0950, Vitis 2023.2.1, AI 3.5 Models for DPUCV2DX8G
AMD Versal™ AI Edge Evalboard with VE2302 device, 8 GB DDR4 SDRAM, 15 x12 cm
https://shop.trenz-electronic.de/en/TE0950-03-EGBE21C-AMD-Versal-AI-Edge-Evalboard-with-VE2302-device-8-GB-DDR4-SDRAM-15-x12-cm?path=Trenz_Electronic/Development_Boards/TE0950/Reference_Design/2023.2/test_board
TE0950 Basic Linux Example archive for Vivado 2023.2.1, build 4, from 31.05.2024 - Versal device
https://shop.trenz-electronic.de/trenzdownloads/Trenz_Electronic/Development_Boards/TE0950/Reference_Design/2023.2/test_board/
TE0950-test_board-vivado_2023.2-build_4_20240531092954.zip
TE0950 Basic Linux Example archive for Vivado 2023.2.1, build 4, from 31.05.2024 - Artix device
https://shop.trenz-electronic.de/trenzdownloads/Trenz_Electronic/Development_Boards/TE0950/Reference_Design/2023.2/test_board/
TE0950-test_board_artix-vivado_2023.2-build_4_20240531084104.zip
DPUCV2DX8G for Versal Adaptive SoCs Product Guide (PG425)
https://docs.amd.com/r/en-US/pg338-dpu/Introduction?tocId=3xsG16y_QFTWvAJKHbisEw

OIder Vitis AI Versions

Not available

Trenz Electronic Documentation

Requirements

Install packages

Install Docker

Install Docker Images for Vitis AI 3.5

Vitis AI 3.5 Model Downloads

Python Downloader

Arch.json File

Compile Pytorch Models

Compile Pytorch Model for AMD DPUCV2DX8G

Compile Other Pytorch Models for AMD DPUCV2DX8G

Sample Vitis AI 3.5 applications

Compile Tensorflow Models

Compile Tensorflow2 Models

Test Sample Vitis AI 3.5 Applications

Install Archives on Systems with Petalinux 2023.2

Compile Vitis AI 3.5 Applications on TE0950 Board

Test Vitis AI 3.5 Sample Applications on TE0950

Export Path to the DPU

Run test_free_run app. and face_mask app. in parallel

Other Vitis AI 3.5 Applications on TE0950 with AMD DPUCV2DX8G

Measured Vitis AI 3.5 Performance on TE0950 with AMD DPUCV2DX8G

TE00950 board with AMD DPUCV2DX8G

References

OIder Vitis AI Versions

Table of contents

Trenz Electronic Documentation

TE0950 AI 3.5 Models for DPUCV2DX8G

Requirements

Install packages

Install Docker

Install Docker Images for Vitis AI 3.5

Vitis AI 3.5 Model Downloads

Python Downloader

Arch.json File

Compile Pytorch Models

Compile Pytorch Model for AMD DPUCV2DX8G

Compile Other Pytorch Models for AMD DPUCV2DX8G

Sample Vitis AI 3.5 applications

Compile Tensorflow Models

Compile Tensorflow2 Models

Test Sample Vitis AI 3.5 Applications

Install Archives on Systems with Petalinux 2023.2

Compile Vitis AI 3.5 Applications on TE0950 Board

Test Vitis AI 3.5 Sample Applications on TE0950

Export Path to the DPU

Run test_free_run app. and face_mask app. in parallel

Other Vitis AI 3.5 Applications on TE0950 with AMD DPUCV2DX8G

Measured Vitis AI 3.5 Performance on TE0950 with AMD DPUCV2DX8G

TE00950 board with AMD DPUCV2DX8G

References

OIder Vitis AI Versions

Table of contents