1.Train YOLOv2 object detector from scratch with Tensorflow.

Prepare two files:

data.csv (three columns: filenames, rois, classes - each row contains image filepath, list of rois (each [x,y,w,h]), list of classes) and anchors.txt (each row contains width and height of one anchor).

python make_tfrecord.py
python train.py
python eval.py

2.Evaluate YOLOv2 model trained with COCO dataset using Tensorflow. Conversion from Darknet to Tensorflow framework done with darkflow project.