Inference of YOLOv8 models exported as onnx models with C++ Builder (Embarcadero) -- 2

The Embarcadero C++ Builder version 11.2 should be used for inference models developed with YOLOv8 which are exported as onnx models.

1) Open sample01.ipynb (based on the tutorial.ipynb from YOLOv8) in Colab and run

2) After running through sample01.ipynb there are
a) 3 sample models:
- model_det Detection model model_det started with YOLO('yolov8n.pt') # load a pretrained YOLOv8n detection model
- model_seg Segmentation model model_seg started with YOLO('yolov8n-seg.pt') # load a pretrained YOLOv8n segmentation model
- model_cls Classification model model_cls started with YOLO('yolov8n-cls.pt') # load a pretrained YOLOv8n classification model
b) sample results for usage of the models with the sample images (included in the outputs)

3) Each model should be exported as onnx and downloaded for inference calculation on a Windows 10 system using the CPU (no GPU!). Colab notebook (*.ipynb) with export code of models should be given.

4) The exported onnx models have to be used in a C++ Builder project. The basic framework is given in InferenceModels01.zip. Calculation of the sample bus.jpg (see sample01.ipynb) must have same results as in sample01.ipynb.
For the C++ Builder project the following requirements have to be fulfilled:
a) Only direct use of *.c, *.cpp, *.h and *.hpp files, no installation of components. DLLs can be used under certain circumstances, but only after consultation.
b) All used files have to have MIT or similar licence so that they can be used for commercial usage without showing source code.
c) Project works with (and any combination of them)
- configuration debug/release
- platform windows 32 bit/64 bit
- option C++ linker -> link with dynamic RTL true/false
- option packages -> linking with run time packages true/false


Prerequisites:
- access to Embarcadero C++ Builder (current version 11.2; maybe a free test version is available)
- experience with Embarcadero C++ Builder and Delphi
- knowledge and access to Colab


Remarks
- If possible, you are welcome to use https://github.com/microsoft/onnxruntime
- documented code has to be delivered
- once more: results must be the same as with original (not exported) models
- It is not possible to get access to my pc
- Payment will only be done if all parts are done sucessfully.


Attached files
- sample01.ipynb File/sample for Colab usage
- InferenceModels01.zip basic framework C++ Builder project


TASKS
Above 3) and 4) have to be done. Of course, I need the source and have to be able to build and run it on my PC (Windows 10).