Hey I've been working on making a traffic surveillance system for my client. I tried to integrate your UltimateALPR in my project, I wrote reco2.py which simply use rtsp and apply your model on it, a modified copy of recognizer.py.
The file runs very well but for some reason its performance is really poor because itsn't using NVIDIA RTX 4070 I have.
I did install 11.8 and respected tensorRT& Cudnn mention in Readme files but its not utilizing my GPU.
Following is the message I got.
```
message: [UltOcrRecognizer] Your GPU supports CUDA and your OCR module isn't GPGPU optimized to run on TensorRT.
ompVGpu] === Number of CUDA devices: 1 ===
*[COMPV INFO]: [CompVCamera] Initializing [camera] module (v 1.0.0)...
*[COMPV INFO]: [CompVDrawing] Initializing [drawing] module (v 1.0.0)...
*[COMPV INFO]: [CompVDrawing] /!\ No jpeg decoder found
*[COMPV INFO]: [CompVDrawing] Drawing module initialized
*[COMPV INFO]: [CompVGpu] GPU enabled: true
*[COMPV INFO]: /!\ Code in file '/home/ultimate/ultimateBase/lib/source/ultimate_base_engine.cxx' in function 'init' starting at line #75: Not optimized for GPU -> GPGPU computing not enabled or deactivated
*[COMPV INFO]: [UltAlprSdkEnginePrivate]Is CUDA correctly configured: Yes
*[COMPV INFO]: [UltAlprSdkEnginePrivate]IC took 39 millis
*[COMPV INFO]: [CompVCpu] Enabling asm code
*[COMPV INFO]: [CompVCpu] Enabling intrinsic code
*[COMPV INFO]: [UltAlprSdkEnginePrivate]Both TensorRT and OpenVINO are enabled. Disabling OpenVINO...
*[COMPV INFO]: [UltAlprSdkEnginePrivate]recogn_tf_num_threads: 32, acceleration backend: OpenVino,TensorRT,
*[COMPV INFO]: [UltAlprRecognizer] TensorRT enabled but plugin not loaded, do not create the plugin.
````
I really like the inference speed the ultimateSDK has, and I had really like to bring this to my client's software. And I had more than willing to buy the license but I'm not getting the inference speed as its been promised from ultimateSDK. Therefore I had like to request a small support on how do I configure it to make it work my GPU.
Looking forward to your reply.
Regards,
Hassan Revel