• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Video Frame Interpolation

Student: Solovev Ilya

Supervisor: Anton Konushin

Faculty: Faculty of Computer Science

Educational Programme: Applied Mathematics and Information Science (Bachelor)

Final Grade: 9

Year of Graduation: 2018

Video frame interpolation is a scope of methods which allow increasing the frame rate. In other words, in a video represented by a sequence of images, these algorithms predict interimediate ones. In this research we explore existing approaches which solve the interpolation problem, considering their advantages and disadvantages with the focus on methods which use optical flow and adaptive convolutions. An approach of adaptive separable convolutions (SepConv), which showed state-of-the-art results, allows obtaining an output image in one pass via neural network by predicting filters for convolving two input frames. While most approaches concentrate on generating one in-between frame, this paper proposes an improvement of the SepConv method, which helps to predict an intermediate frame at a given point by adding time information into featuremaps of a neural network. Implicit insertion of a timestamp in this form was not previously presented in video frame interpolation papers. Obtained results of the algorithm are illustrated with synthetic and real data. Moreover, we propose a number of modifications for the SepConv method by adding pre-defined filters that allow reconstructing more complex frame structures. Also, possible improvements of the proposed method are presented in the study. Keywords: video frame interpolation, optical flow, convolutional neural networks, residual connections, adaptive separable convolutions, perceptual loss.

Full text (added May 21, 2018)

Student Theses at HSE must be completed in accordance with the University Rules and regulations specified by each educational programme.

Summaries of all theses must be published and made freely available on the HSE website.

The full text of a thesis can be published in open access on the HSE website only if the authoring student (copyright holder) agrees, or, if the thesis was written by a team of students, if all the co-authors (copyright holders) agree. After a thesis is published on the HSE website, it obtains the status of an online publication.

Student theses are objects of copyright and their use is subject to limitations in accordance with the Russian Federation’s law on intellectual property.

In the event that a thesis is quoted or otherwise used, reference to the author’s name and the source of quotation is required.

Search all student theses