Python Video Text Eraser
Budget: $250 – $750 USD
I need a clean, well-commented Python 3 script that can scan any standard video file and remove every kind of on-screen text—not just subtitles, watermarks, or annotations, but any other text that may pop up throughout the footage.
Please use the most up to date technology for this. Maybe a good video in-painting models, OCR-guided masking—whatever combination reliably wipes the text while preserving the underlying picture, audio, and frame rate.
Please make the tool:
• Format-agnostic on input (it must run at least on MP4, with AVI and MOV welcome as a bonus).
• Fully offline, depending only on open-source libraries installable via pip or conda.
• Executable from the command line with a single input and output argument, optional batch folder mode is a plus.
• Smart enough to keep non-text elements untouched so there are no blurred blotches after processing.
Deliverables
1. text_remover.py with clear dependency list and inline comments.
2. A short README explaining installation, usage, and any tunable parameters.
3. One sample before/after clip (5–10 s) demonstrating that the script truly removes arbitrary text without degrading the video.
Acceptance criteria
• Running `python text_remover.py input.mp4 output.mp4` produces a video where visible text is gone across all frames.
• Visual artifacts are minimal—no noticeable flicker or color shifts at normal playback speed.
• Script completes on a mid-range GPU or CPU in reasonable time for a 2-minute 1080p clip.
If you’ve built something similar or have a clever deep-learning trick up your sleeve, let me know—speed and quality both matter here.
Please use the most up to date technology for this. Maybe a good video in-painting models, OCR-guided masking—whatever combination reliably wipes the text while preserving the underlying picture, audio, and frame rate.
Please make the tool:
• Format-agnostic on input (it must run at least on MP4, with AVI and MOV welcome as a bonus).
• Fully offline, depending only on open-source libraries installable via pip or conda.
• Executable from the command line with a single input and output argument, optional batch folder mode is a plus.
• Smart enough to keep non-text elements untouched so there are no blurred blotches after processing.
Deliverables
1. text_remover.py with clear dependency list and inline comments.
2. A short README explaining installation, usage, and any tunable parameters.
3. One sample before/after clip (5–10 s) demonstrating that the script truly removes arbitrary text without degrading the video.
Acceptance criteria
• Running `python text_remover.py input.mp4 output.mp4` produces a video where visible text is gone across all frames.
• Visual artifacts are minimal—no noticeable flicker or color shifts at normal playback speed.
• Script completes on a mid-range GPU or CPU in reasonable time for a 2-minute 1080p clip.
If you’ve built something similar or have a clever deep-learning trick up your sleeve, let me know—speed and quality both matter here.