chiral-carbon / frame-pred-segment

Video future frame prediction and segmenting the final predicted frame using SimVP and U-Net.

Project Description

Problem Statement:

To predict 22nd frame from given 11 frames of a video and perform semantic segmentation on the 22nd image.

Folders Description

Model 1 - SimVP

• This folder contains code for first part of the project which is Future frame prediction.
• Refer the README.md inside this folder to execute the code.

Model 2 - U-Net

• This folder contains code for second part of the project which is Semantic segmentation and masks generation.
• Refer the README.md inside this folder to execute the code.

Final Pipeline

• This folder contains code for third part of the project which is Pipeline which takes the saved model of part 1 and part 2 and hidden datset and generates the required final output numpy array.
• Refer the README.md inside this folder to execute the code.

Sequence to Reproduce

1. Run code in Model 1-SimVP
2. Run code in Model 2-UNet
3. Use the trained models generated in part 1 and part 2 along with the hidden dataset to work with Final Pipeline