project.single_upload() of image and corresponding annotation file in yolov8 txt format uploads to the project but annotation is recognized as a polygon instead of keypoints. Predictions were generated by the model in the project using the python roboflow python api. Prediction results were then translated to yolov8 txt annotation format.
the resulting uploaded image and the annotation, see: digital-gauge Dataset and Pre-Trained Model by Alexander Jaworowski
used the same image as digital-gauge Dataset and Pre-Trained Model by Alexander Jaworowski. This image was manually annotated in the roboflow web interface. So i was able to compare that my yolov8 annotation file (txt) was correct before uploading (since i can compare with the dataset label for that image by downloading the dataset in yolov8 format).
annotation file test_0001.txt content is:
0 0.4584199584199584 0.31728045325779036 0.15176715176715178 0.4702549575070821 0.5218295218295218 0.5410764872521246 2 0.39293139293139295 0.0906515580736544 2