How to create a program on face detection with the help of training data obtained from a confidence map created in an empty environment to find the best position for face detection?
I just need some ideas on how to implement this face detection on a gazebo world with people in a particular scenario. the face detection used is haar cascade frontal face detector. Everything's working fine including autonomous navigation. the objective is to write a program to move to the best position in the scenario with respect to the person where the best position is determined having maximum probability of determining the face. if this is an occluded environment it need to switch to best position within the constraints where the face is best detected.
For this I created an empty world with a single person, done executing face detection and determine those positions having the maximum probability of determining face and created a heatmap. this is my confidence map currently the coordinates created with respect to the origin of the map. Based on this map i need to superimpose this into my current scenario to execute the above mentioned task to get the results for best viewpoint. TIAGo robot is used for this purpose.
my current un-clarified doubts are: 1. How will i superimpose the data onto the scenario. I can shift the readings from origin to the person perspective but this might get cumbersome as it will change the x and y coordinates and its orientation since it is with respect to the person and will change with his orientation. 2.if that's the case how to read the position and orientation of the person viewing? Please give me some insights into this as I'm quite new to ROS and python programming. 3. is it possible to perform relative coordinate with respect to the person. cause the office world in gazebo consists of 5-6 people and the coordinate needs to shift to each person once the robot reaches the person who's face detection needs to be performed. If so please give me some insights and the function that is required to perform this. the readings on the confidence map will change accordingly with respect to the person.
Kindly just give some suggestions, ideas or functions that executes relative coordinate with respect to object.
Thank you.
