# Deepstream 啟動 ## Check List pth to onnx - [ ] data_wav - [ ] sox wav to audio_16k.wav - [ ] ast_models.py->forward->normalize (x - (mean)) / (std * 2) - [ ] best_audio_model.pth (copy) - [ ] model_3classes_16k.onnx deepstream with onnx - [ ] model_3classes_16k.onnx_b1_gpu0_fp32.engine (delete) - [ ] model_3classes_16k.onnx (copy) - [ ] audio_16k.wav (copy) - [ ] ds_log_result.txt export LD_LIBRARY_PATH=//opt/nvidia/deepstream/deepstream-6.2/lib/:$LD_LIBRARY_PATH ## Linux (ubuntu) 清除屏幕 ``` clear ``` ## 啟動 docker 鏡像資料夾 /mnt/c/deepstream : /tmp/deepstream ``` sudo service docker restart sudo docker run -it --rm --net=host --runtime nvidia -e DISPLAY=$DISPLAY -w /mnt/c/deepstream -v /mnt/c/deepstream:/tmp/deepstream nvcr.io/nvidia/deepstream:6.2-samples ``` # 強制覆蓋複製檔案 (輸出多類別各別機率) ``` cp -f /tmp/deepstream/sample_codes/sources/ITRIAudioClassifier/libnvds_infercustomparser.so /opt/nvidia/deepstream/deepstream/lib/libnvds_infercustomparser.so ``` custom-lib-path=/opt/nvidia/deepstream/deepstream/lib/libnvds_infercustomparser.so 改為 custom-lib-path=libnvds_infercustomparser.so ## ITRI path ``` cd /tmp/deepstream/sample_codes/sources/ITRIAudioClassifier deepstream-audio -c ds_audio_config.txt deepstream-audio -c ds_audio_config.txt > ../../samples/ds_log_result.txt ``` # 強制覆蓋複製檔案 (輸出最大類別與機率) ``` cp -f /tmp/deepstream/sample_codes/sources/ITRIAudioClassifier/deepstream-audio /usr/bin/deepstream-audio ``` ## find deepstream-audio 如何找檔案 ``` whereis deepstream-audio ``` 系統回傳 deepstream-audio: /usr/bin/deepstream-audio ## 複製 deepstream-audio ``` cp /usr/bin/deepstream-audio /tmp/deepstream/deepstream-audio ``` ## SOX 44k to 16k sox cry_44k *.wav -r 16000 cry_16k.wav 命令串接所有音檔 sox *.wav long.wav ## 廠商範例 ``` cd /tmp/deepstream/SONYCAudioClassifier/ deepstream-audio -c ds_audio_sonyc_test_config_itri.txt ``` log.txt ``` > /ds_log_result.txt ``` 官網(複製檔案)範例 (cd /tmp/deepstream/ds) ``` sudo docker run -it --rm --net=host --runtime nvidia -e DISPLAY=$DISPLAY -w /mnt/c/deepstream -v /mnt/c/deepstream:/tmp/deepstream nvcr.io/nvidia/deepstream:6.2-samples cd /tmp/deepstream/ds/opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-audio/configs deepstream-audio -c ds_audio_sonyc_test_config.txt deepstream-audio -c ds_audio_sonyc_test_config_car.txt ``` docker start (官網範例) ``` docker run --gpus "device=0" -it --rm -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=$DISPLAY -e CUDA_CACHE_DISABLE=0 nvcr.io/nvidia/deepstream:6.2-samples cd /opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-audio/configs deepstream-audio -c ds_audio_sonyc_test_config.txt ``` # Configuration 設定參考 ## ds_audio_sonyc_test_config.txt ### 多資料來源設定 source來源預設是給兩個, 所以跑的結果會有 source_id:[0] 或 source_id:[1] 只要把[source1]相關設定拿掉，就能只跑一種來源 ``` [source0] enable=1 #Type - 2=URI type=6 uri=./samples/sonyc_mixed_audio.wav num-sources=1 gpu-id=0 砍掉source1即可 [source1] enable=1 #Type - 2=URI type=6 uri=./samples/sonyc_mixed_audio.wav num-sources=1 gpu-id=0 ``` ### 測試集loop迴圈設定 停止測試集無限迴圈測試 file-loop=1 ``` 預設: [tests] file-loop=0 修改: [tests] file-loop=1 ``` ### threshold [property] classifier-threshold = 0.8 ## 官網連結 [Configuration Groups](https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_ref_app_deepstream.html) [Gst-nvinferaudio](https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvinferaudio.html) # Parser CUDA Makefile 測試 顯示 CUDA Runtime Version: 11.8 ``` deepstream-app --version-all ``` 執行 Makefile ``` cd /opt/nvidia/deepstream/deepstream/sources/SONYCAudioClassifier/gstnvinferaudio_custom_parser CUDA_VER=11.8 make ``` error code ``` /usr/local/cuda-11.8/include/cuda_runtime_api.h:147:10: fatal error: crt/host_defines.h: No such file or directory 147 | #include "crt/host_defines.h" compilation terminated. make: *** [Makefile:41: libnvds_infer_custom_parser_audio.so] Error 1 ``` 嘗試解決 ``` cd /usr/local/cuda-11.8/include/ export PATH=/usr/local/cuda-11.8/bin:$PATH export LD_LIBRARY_PATH=/usr/local/cuda-11.8/lib64:$LD_LIBRARY_PATH export PATH=/usr/local/cuda-11.8/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-11.8/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}} export CUDA_HOME=$CUDA_HOME:/usr/local/cuda-11.8 ```