egesabanci / ex-machina-conv2d

:robot: Applying 2D convolutions with extracted frames of the sample scene from Ex Machina movie

Overview

Applying 2D convolutions with extracted frames of the sample scene from Ex Machina movie for creating something artistic. Input images are 14 extracted frames from the sample scene of the Ex-machina movie. After the frame extraction, random 2D Convolutions has been applied. Then, these 14 images gathered back together for creating the final GIF.

Output

Network Architecture

          CompositeVGG19 - type: Sequential (Pretrained VGG19 based model)
                [first 3 layers with input shape of (720, 1280, 3)]
                                   (imagenet)
                                   ----------
                                        |
                                        |
                        Conv2D (with kernel size of (1, 1))
                                        |
                                        |
                                   ----------
                  Conv2D (with kernel size of (1, 1) and 3 filters)
                                        |
                                        |
                                   ----------
                        Conv2D (with kernel size of (1, 1))
                                        |
                                        |
                                   ----------
                      BatchNormalization (with default values)

Lives on the Blockchain! ⛓️

This project's output also minted and listed on OpenSea

Get the item!