美文网首页
视频生成小综述起稿

视频生成小综述起稿

作者: Feather轻飞 | 来源:发表于2018-05-06 22:05 被阅读0次

    Year 2018

    March

    1. Probabilistic Video Generation using Holistic Attribute Control https://arxiv.org/pdf/1803.08085.pdf

       a. Videos express highly structured spatio-temporal patterns of visual data. two factors:

            (i) temporally invariant (e.g., person identity), or slowly varying (e.g., activity), attribute-induced appearance, encoding the persistent content of each frame

            (ii) an interframe motion or scene dynamics (e.g., encoding evolution of the person ex- ecuting the action).

       b. VideoVAE

           video generation + future prediction.

           generates a video (short clip) by:

               decoding samples sequentially drawn from a latent space distribution into full video frames.

                  -VAE: encoding/decoding frames into/from the latent space

                  -RNN: model the dynamics in the latent space.    

            improve the video generation consistency through temporally-conditional sampling and quality

                  -structuring the latent space with attribute controls

                  -ensuring that attributes can be both inferred and conditioned on during learning/generation

    2.Learning to Generate Time-Lapse Videos Using Multi-Stage Dynamic Generative Adversarial Networks

    3.Every Smile is Unique: Landmark-Guided Diverse Smile Generation 

    Year 2017

    -By the Way

     I like this stanford homework paper http://cs231n.stanford.edu/reports/2017/pdfs/323.pdf

    1. Dynamics Transfer GAN: Generating Video by Transferring Arbitrary Temporal Dynamics from a Source Video to a Single Target Image

    -spatial constructs <---- target image; dynamics <------source video sequence

     To preserve the spatial construct of the target image:

                 - the appearance of the source video sequence is suppressed

                 - only the dynamics are obtained before being imposed onto the target image.  (using the proposed appearance suppressed dynamics feature.)

     the spatial and temporal consistencies are verified via two discriminator networks.  

                 - discriminator A validates the fidelity of the generated frames appearance,

                 -  B validates the dynamic consistency of the generated video sequence.

    Results:

                 - successfully transferred arbitrary dynamics of the source video sequence onto a target image

                 - maintained the spatial constructs (appearance) of the target image while generating spatially and temporally consistent video sequences.

    Note: It is ### everything (Literature Review in its intro) because it is quite new.

    2. Deep Video Generation, Prediction and Completion of Human Action Sequences https://arxiv.org/pdf/1711.08682.pdf


    3. Video Generation from Text https://arxiv.org/pdf/1710.00421.pdf

    -Hybrid VAE plus GAN

    -Two parts:

    -Static( Using gist to sketch text-conditioned background color and object layout (LSTM, RNN structure)

    -Dynamic ( A text2Filter. )

    -3.3 Text2Filter

    -Note: Quite compact. Need time to digestilter

    4. Learning to Generate Time-Lapse Videos Using Multi-Stage Dynamic Generative Adversarial Networks

       https://arxiv.org/pdf/1709.07592.pdf

    5. MoCoGAN: Decomposing Motion and Content for Video Generation

       https://arxiv.org/pdf/1707.04993.pdf


    6. To Create What You Tell: Generating Videos from Captions

        https://www.microsoft.com/en-us/research/wp-content/uploads/2017/11/BNI02-panA.pdf


    -Temporal GANs conditioning on Captions, namely TGANs-C

         - transformed into a frame sequence with 3D spatio-temporal convolutions.

          -  GAM evaluation metric ( Section 3.4 Experimental Setting)

    -  Model Architecture

                -3.1.1 Generator

                         -Given a sentence 𝒮, a bi-LSTM is utilized to contextually embed the input word sequence,  + a LSTM- based encoder to obtain the sentence representation S. + concatenated input of the sentence representation S and random noise variable z.synthesize realistic videos with these

                 -3.1.2 The discriminator network 𝐷 includes three discriminators:

                               a.video discriminator classifying realistic videos from generated+ optimizes video-caption matching           

                               b. frame discriminator( between real and fake frames)and aligning frames with the conditioning caption

                               c. motion discriminator emphasizing that the adjacent frames in the generated videos run smoothly

                  -3.1.3 The whole part trained with 3 losses:video-level matching-aware loss, frame-level matching-aware loss and temporal coherence loss

                       .

       Year 2016

    1. Generating Videos with Scene Dynamics

         https://arxiv.org/abs/1609.02612

    - a spatio-temporal convolutional architecture

    - untangles the scene’s foreground from the background.

    - experiments show the model internally learns useful features for recognizing actions with minimal supervision,

    - scene dynamics are a promising signal for representation learning.

    - Slides : https://pdfs.semanticscholar.org/presentation/7188/6726f0a1b4075a7213499f8f25d7c9fb4143.pdf

    相关文章

      网友评论

          本文标题:视频生成小综述起稿

          本文链接:https://www.haomeiwen.com/subject/pkrprftx.html