Train set is the set of data-points on which the model learns/trains, Validation set is used to compare between models and Test set to see how well does our model generalize to unseen input (input which we have not trained upon)
Hyper-parameter is a configurable value which is set before the learning process begins. These hyper-parameter values dictate the behavior of the training algorithm and how it learns the parameters from the data.
Each bounding box will have a score associated (likelihood of the box containing an object). Based on the predictions a precision-recall curve (PR curve) is computed for each class by varying the score threshold. The average precision (AP) is the area under the PR curve. First the AP is computed for each class, and then averaged over the different classes. The end result is the mAP.