如何提高神经网络运行速度，跑的时候只有3帧，太卡了

[ikcg]

我训练了5种训练集，加上颜色识别 传输到单片机上，如何提高运行速度呢？

thresholds = [
(15, 76, 9, 68, -40, 47), #
(30, 100, -64, -8, -32, 32), #
(25, 77, -24, 10, -40, 2), #
(0, 15, 0, 40, -80, -20)] #
A_net = "A_trained.tflite"
A_labels = [line.rstrip('\n') for line in open("a_labels.txt")]
B_net = "B_trained.tflite"
B_labels = [line.rstrip('\n') for line in open("b_labels.txt")]
C_net = "C_trained.tflite"
C_labels = [line.rstrip('\n') for line in open("c_labels.txt")]
D_net = "D_trained.tflite"
D_labels = [line.rstrip('\n') for line in open("d_labels.txt")]
E_net = "E_trained.tflite"
E_labels = [line.rstrip('\n') for line in open("e_labels.txt")]

[kidswong999]

没办法，计算量就是很大。