In this tutorial, we are going to discuss the project named mix and match. The mix and match project is based on a conditional generative model. So, the conditional generative model is used to learn to encode and disentangle background, pose, shape, and texture of objects from real images with minimal supervision for the generation of mixed images. This project is based on FineGAN which is an unconditional generative model. So, the unconditional generative model is used to study the required breakdown and picture generator, and use adversarial mutual picture-code delivery to acquire latent factor encoders. Min and match include bounding boxes to the model background during training but no further monitoring is needed.
The following are the activities that we have done while building a mix and match project.
- a. Accurately disentangle, encode and combine multiple factors for mix and match image generation
- b. Including sketch2color
- c. Cartoon2img
- d. Img2gif
Technologies used in mix and match program:
- a. Linux
- b. Python 3.7 or above
- c. Pytorch 1.3.1
- d. NVIDIA GPU + CUDA CuDNN
Step involved to run this program:
- a. Download the repository and unzip the file.
- b. Download the formatted CUB data (Download Link) and extract downloaded CUB data inside the data directory.
- c. Evaluated the model
- d. You can train your own model
- e. Finally enjoy this program