I am currently searching a neural network that can classify if there is a human in an image or not.
I checked the ImageNet dataset, but the 1000 classes there contain nothing like human or person or such, mostly animals and things.
Segmentation CNNs usually have a lot longer runtime, and all I really need is basically a binary NN, just "Yes = Human in Image", "No = No human in image". Can someone point me to a training set or a pre-trained network?
In Part 1 of this "tutorial/discussion" ther is a good explanation of the problem and a possible solution using OpenCV built in Haar Cascades or HOG for Human Detection (this is the key-word you should use to search info about your task).
They are both really fast compared to NN and require little training.
In Part 2 are explained more modern approaches using Deep Learning. A notable mention would be AlexNet.
At the end there is even a tutorial on Setting up a Basic Human Detector with Tensorflow.
