emmafaye/ssbu-replay-analyzer

Replay Analyzer for Smash Ultimate is for content-creators, shoutcasters, professionals, and casuals seeking to improve their content delivery and reviews of replays

Super Smash Bros. Ultimate - Replay Analyzer

⚠️ Currently a Work-in-Progress and is not ready for offical usage

⭐ this repo for a future release or feel free to assign an open issue to yourself!

Replay Analyzer for Smash Ultimate is for content-creators, shoutcasters, professionals, and casuals seeking to improve their content delivery and reviews of replays. This project was born out of Nintendo's lack of support for a user-friendly replay viewer that allows easy sharing and review.

Features

• Feed-in your recordings from your capture card or stream
• Auto-cut the video into it's relevant matches and sets
• Allows for easy review and sharing of matches through YouTube

ssbu-replay-analyzer's Issues

QoL Features

• Fix logging to output one stream when calling script without stdout
• Fix color in terminal
• Score results.log and average matches
• Tweak parameters for test.sh to get highest score with overall tests
• Fix colors breaking emojis and adding unicode in console output
• Set logging level to debug for file and logging and info for console
• Clean up output image paths to use relative directories instead of absolute
• Output image path for each Found item
• Add python-based progess bar for console as script runs
• Try out different colored logs with progress indicator (https://github.com/kopensource/colored_logs)

Desktop App

Create a desktop app for our python script using QT to create the GUI.

• Research QT and intergration with Python (https://www.qt.io/qt-for-python)
• Create a barebone GUI and desktop app

Region Detection

The ultimate goal is to consistently detect key items like when a game starts and ends with the GO! and GAME! text. In addition the player tags to provide consistent data on matches.

• Region detection is one method of potentially guaranteeing better results

• Add a function to add and store regions with (x,y,w,h) to parse for expected text

Support Video Processing

• Clean-up _parse_text() pre-processing code
• Add option to enable/disable match outputs in _parse_text()
• Create process function that splits logic from main() function
• Refactor main() function to helper functions for various components
• Process Video Capture frames for entire video and extract text
• Refactor processing to accept video or image
• Research capture card device stream and plugging it directly into the algorithm

Optimize Text Recognition

• Remove contours from images to add clarity