Abstract: This disclosure provides a lens assembly that has an optical path and includes a lens element and a light-blocking membrane layer. The lens element has an optical portion, and the optical path passes through the optical portion. The light-blocking membrane layer is coated on the lens element and adjacent to the optical portion. The light-blocking membrane layer has a distal side and a proximal side that is located closer to the optical portion than the distal side. The proximal side includes two extension structures and a recessed structure. Each of the extension structures extends along a direction away from the distal side, and the extension structures are not overlapped with each other in a direction in parallel with the optical path. The recessed structure is connected to the extension structures and recessed along a direction towards the distal side.

Abstract: An object detection method and an object detection system are provided. The object detection method includes the following steps. An object selected range in a current image is filtered out to obtain a background range. An object foreground range and an object background range are obtained from the object selected range. If the number of object foreground moving vectors corresponding to the object foreground range is greater than the number of background moving vectors in the object background range, a foreground moving trend of the object foreground moving vectors corresponding to the object foreground range is calculated. The object selection range of the current image is updated according to the foreground moving trend of the object foreground moving vectors corresponding to the object foreground range or a background moving trend of the background moving vectors corresponding to the background range.

Abstract: An image correction method and an image correction system based on deep learning are provided. The image correction method includes the following steps. An image containing at least one character is received by a deep learning model, and a perspective transformation matrix is generated according to the image. A perspective transformation is performed on the image according to the perspective transformation matrix to obtain a corrected image containing a front view of the at least one character. An optimized corrected image containing the front view of the at least one character is generated according to the image. An optimized perspective transformation matrix corresponding to the image and the optimized corrected image is obtained. A loss value between the optimized perspective transformation matrix and the perspective transformation matrix is calculated. The deep learning model is updated using the loss value.