h264_decode (GPT Gen)

Overview

The h264_decode repository is designed to explore and test the decoding of H.264 bitstreams using Python, specifically focusing on intra-frame prediction techniques. This project originates from experiments with the JM (Joint Model) reference software version 16.1, tailored for decoding H.264 bitstreams.

Current Functionality

As of now, the repository contains preliminary Python scripts that facilitate the decoding process with an emphasis on intra-frame prediction. The current implementation is experimental and aims to decode the bitstreams to raw traces, primarily focusing on intra-prediction modes.

Problem Statement

The intra-frame prediction currently does not perform adequately:

• The prediction mode for blocks of size 8x8 and 4x4 is not present after decoding to traces.
• The Inverse Discrete Cosine Transform (IDCT) results are suboptimal for block sizes 8x8 and 4x4 when compared to a 16x16 block. Notably, IDCT is only effectively applied in a 4x4 area, regardless of the original block size.

Repository Structure

• REDS_264_bitstream_OUTPUT/ - Output directory for bitstream operations.
• REDS_264_trace_OUTPUT/ - Output directory for trace operations.
• H264_OUTPUT/ - General output directory for H.264 decoding.
• jm_16.1/ - Contains JM reference software version 16.1 used for generating test bitstreams.
• scripts/ - Contains Python scripts for the conversion and processing tasks:
  • p1_pngs2h264.py - Converts PNG images to H.264 bitstream.
  • p2_h264_to_bitstream.py - Extracts bitstream from H.264 video.
  • p3_bitstream_to_traces.py - Converts bitstream to trace files for analysis.
  • png2h264_intra_only_parallel.py - Handles intra-frame prediction in parallel.
• i_frame_decoder.py - Script focused on intra-frame decoding.
• reds_264_decode.ipynb - Jupyter notebook detailing the decoding process and issues.

Getting Started

To get started with this project:

1. Clone the repository.
2. Ensure Python 3.x is installed on your machine.
3. Install required dependencies: pip install -r requirements.txt (to be added).
4. Explore the reds_264_decode.ipynb for a step-by-step guide on the current decoding process and intra-frame prediction challenges.

How to Contribute

Contributions are welcome, especially from those with expertise in video compression and Python programming. To contribute:

1. Fork the repository.
2. Create a new branch for your feature or fix.
3. Develop your feature or fix.
4. Submit a pull request against the main branch.

Seeking Help

We are specifically seeking help in improving the intra-frame prediction algorithms and enhancing the IDCT results for various block sizes. Any insights, code enhancements, or suggestions in these areas would be greatly appreciated.