Step 1:

install jtop by running this command "sudo pip3 install -U jetson-stats" Run "jtop" Go to page 6. Type 'S' then 'E' then select 'manual' in the profiles section.

Step 2:

Go to 'Info' page in jtop, check if opencv with CUDA is available. If not, follow the link below: https://www.youtube.com/watch?v=art0-99fFa8

Step 3:

Copy Makefile and YOLOv8-Jetson.py to root directory

Run "cd ~" then "make install"

Step 4:

Run "cd DeepStream-Yolo" then run "git checkout 68f762d5bdeae7ac3458529bfe6fed72714336ca" then run "cp ~/DeepStream-Yolo/utils/gen_wts_yoloV8.py ~/ultralytics"

Step 5:

Run "cd ~/ultralytics" Edit the gen_wts_yoloV8.py Remove cd ~/ultralytics from line 48, if it doesn't exist move forward Remove ".yolo" from line 5 and line 6 Edit line 287 and 288 to add yolov8s

Step 6:

Copy ultralytics's Makefile to ultralytics folder, then run "cd ~/ultralytics" then run "make run"

Step 7:

Copy DeepStream-Yolo's Makefile to DeepStream-Yolo folder, then run "cd ~/DeepStream-Yolo" then run "make run"

Step 8:

Copy config_infer_primary_yoloV8.txt to DeepStream-Yolo/

Step 9:

Copy deepstream_app_config.txt to DeepStream-Yolo/

Step 10:

Copy labels.txt to DeepStream-Yolo/

Step 11:

Testing

deepstream-app -c deepstream_app_config.txt

NOTE:

Nvidia Guide Link to work with configuration: https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_ref_app_deepstream.html