Week Project-YOLO V2 Testing & PR curve

In this section we'd like to implement multiclass object detection and evaluate trained YOLOv2 model.

BCCD datasets

Reference: Datasets download

BCCD dataset contains three kind of labels: Red Blood Cell, White Blood Cell, and Platelets. In the directory there are 410 images and corresponding annotation files, and we divided the dataset into training set and testing set with 200 images and 210 images respectively. However, in two or three annotation files(.xml), '.jpg' is missing between the opening and closing tag of 'filename' that would cause errors, just adding '.jpg' manually can fix it.

Tensor Board

There are total 25 epochs and it stopped because we applied 'EarlyStopping' callback function during my training process. 'EarlyStopping' stops training when the validation data set grows several times.

Test

AP=80%