I would like to use a deep learning approach to detect people in videos. I have found some freely accessible implementations like Human Segementation with Pytorch or BodyPix / DeepLab / Pixellib with Tensorflow. They all work well, but with many it happens that, for example, half hand is not detected or if a person is sitting in the picture only the legs and the head are detected. Are there other approaches to detect people who are freely accessible or is that state-of-the-art? I had imagined such problems have been solved, but I don't know so much about it. Thanks for your answers.
You can use any state of the art object detection such as DETR, EfficientDET, YOLO etc. since they have an object 'person' in them. But, if the density of people is high in the images, then, you should look at 'Finding Tiny Faces' which is good for that problem (this will help you detect faces only, think group photos). If you want to find the whole person even if they are occluded, then, it will be difficult (you can then think of segmentation techniques and then, try to find out the people but it will be more taxing).