A binocular-vision-based method for tracking fruit space attitude and fruit space motion, the method comprising: establishing a connected base coordinate system by taking a junction of a fruit and a fruit stem as an origin; statically photographing a feature point on the surface of the fruit and a point of the connected base coordinate system established at the junction of the fruit and the fruit stem; storing a photographed image; acquiring an inherent relationship between the feature point and the connected base coordinate system; photographing dynamic motion of the fruit; acquiring absolute coordinates of the feature point on the surface of the fruit; calculating, according to the inherent relationship between the feature point and the connected base coordinate system, absolute coordinates of a point of the connected base coordinate system at each moment corresponding to each frame of image; and respectively calculating the displacement, instantaneous speed and instantaneous acceleration of the fruit, calculating swing angular displacement and swing angular acceleration of the fruit, and calculating a fruit torsion angular speed and a fruit torsion angular acceleration at the moment t. The study of a fruit motion state in the field of forest fruit harvest through vibration is performed, so that the motion of fruits can be better tracked.