Video holds significance in computer graphics applications. Because of the heterogeneous of digital devices, retargeting videos becomes an essential function to enhance user viewing experience in such applications. In the research of video retargeting, preserving the relevant visual content in videos, avoiding flicking, and processing time are the vital challenges. Extending image retargeting techniques to the video domain is challenging due to the high running time. Prior work of video retargeting mainly utilizes time-consuming preprocessing to analyze frames. Plus, being tolerant of different video content, avoiding important objects from shrinking, and the ability to play with arbitrary ratios are the limitations that need to be resolved in these systems requiring investigation. In this paper, we present an end-to-end RETVI method to retarget videos to arbitrary aspect ratios. We eliminate the computational bottleneck in the conventional approaches by designing RETVI with two modules, content feature analyzer (CFA) and adaptive deforming estimator (ADE). The extensive experiments and evaluations show that our system outperforms previous work in quality and running time.

Keywords -- video retargeting, RETVI, analyze video, deforming, grid movement, pixel movement

We compare our RETVI with a conventional video retargeting method, [Lin et al]. Two major drawbacks could be obviously seen in Lin's results here. First, the video is not stable. Flickering artifact occurs seriously in the background. Second, the shapes of main objects in video are distorted or shrinking. Our RETVI can produce results without these phenomena.

We compare our RETVI with a commercial application, Adobe, on the performance of resizing video to 9:16 aspect ratio. This is the case that the video is with multiple moving object and camera-move. As Adobe uses a window of size 9:16 is allocated at the middle of video and they manually crop two sides of video, they can keep the ratio of object in the same as in the input video The shape of objects may be smaller than those by Adobe, but it can capture the tasteful moment on each frame and appear visually pleasing.

Alike video content, our approach also performs well on different image contents, i.e., portrait image, shaped object, multiple objects, line structure. Images with reflection symmetry are challenging when retargeted by seam carving operator. The image in the last row consists of line structure and symmetrical structure. Our RETVI still works well and produces appealing results without distortion.

Here we further demonstrate the ability of our method via visual comparison with four recent state-of-the-art image retargeting systems, WSSDCNN, SAMIR, grid encoding model, and Cycle-IR.