train_updateNet.py，关于训练部分的问题！

[renlicheng]

trafficstars

dataram = dict() tem0_path = '/home/lichao/projects/DaSiamRPN/code/templates' tem_path = '/home/lichao/projects/DaSiamRPN/code/templates_0_0'

dataram['template0'] = np.load(join(tem0_path,'template0.npy')) dataram['template'] = np.load(join(tem_path,'template.npy')) dataram['templatei'] = np.load(join(tem_path,'templatei.npy'))

dataram['pre'] = np.load(join(tem_path,'pre.npy')) dataram['gt'] = np.load(join(tem_path,'gt.npy')) dataram['init0'] = np.load(join(tem_path,'init0.npy')) dataram['train'] = np.arange(len(dataram['gt']), dtype=np.int)

你好 这部分代码是什么意思呢，这几个.npy文件哪里可以找到呢

[zhanglichao]

你好 They are the templates for training UpdateNet and along with the corresponding labels, i.e. initial, accumulated, current and real, as in the pipeline Figure of tracker (three inputs and one finally compared one).
They are generated by simulating the real tracking.

[Rongmiq]

请问.npy文件怎么生成？