The invention discloses a warehouse signboard identification method based on an image convolution neural network technology. The method mainly comprises the following steps that the interesting partsof pictures are extracted from a photo stream, and photos are zoomed and converted into photos of the same size; signboard photos are detected on the basis of an optical flow method, and when it is detected that the photos change, photos obtained at the moment of changing and the photos of the previous frame and the next frame are extracted; the three photos serve as input, photo computing is conducted by means of a convolution neural network, and finally changes of signboards are judged. Accordingly, the photos are zoomed to reach the fixed size by extracting the area of interest in signboardscene photos, changes of the signboards are detected, block processing is conducted by means of a convolutional neural algorithm, the problems that in large storage, the number of the signboards is large, and scenes are complex, so that detection is difficult are avoided, and the test precision is greatly improved. The method can be widely applied to the field of storage management.