Online video analysis method and system, medium and equipment

The invention discloses an online video analysis method and system, and the method comprises the following steps: determining a video frame difference threshold value for triggering edge target recognition in a next time slot through employing an online learning algorithm according to the recognition accuracy of a previous time slot and the network condition of a current moment; before each frame of picture is processed, calculating the difference between the frame and the previous frame, and comparing the difference with a previously determined threshold value to determine whether the current frame is at a device end or is transmitted to an edge for processing; if the object is processed at the equipment end, calculating an average motion vector by using an optical flow method, and simply tracking the object; and if the image is processed at the edge end, performing target identification on the whole image by using a convolutional neural network. By utilizing the method, the setting of the threshold value ca