['c0_safe_driving', 'c1_texting_right', 'c2_talking_phone_right', 'c3_texting_left', 'c4_talking_phone_left', 'c5_operating_the_radio', 'c6_drinking', 'c7_reaching_behind', 'c8_hair_and_makeup', 'c9_talking_to_passenger']
1.Apply a better model (MobileNet)
2.Increase the size of dataset.