Real-time violent action detection using computer vision
Standard YOLO + LSTM pipeline for violence detection, nothing novel beyond the specific use case.

Another face blurring service competing with Remove.bg and Adobe when local tools already exist.
Content creators and journalists needing to anonymize footage before publishing
Remove.bg · Adobe Express · ObscuraCam
First time poster here, but I feel that my product is finally ready to showcase to HN. Me and my team built an amazing privacy tool that uses computer vision to analyze, detect, and hide sensitive data in photos or videos. It detects the identifiable visual data (faces, license plates, tattoos, phone numbers, etc.) and applies a blur effect over them to remove them from sight automatically!
Right now it only works on faces, we are working to add license plates but the biggest milestone for us has been video support. We just added it and even though it is getting better and improving it is a huge step forward, as it is about 1000x harder than processing an image.
We are currently searching for concrete PMF with business and creator niches, but we would love any feedback the community can provide!
We made it 100% anonymous, with no personal data required for signup, in memory processing only, and zero photos and videos stored! Privacy is our focus. Just create a password and your account will be created!
Standard YOLO + LSTM pipeline for violence detection, nothing novel beyond the specific use case.
Pulse detection from video is technically clever; zero practical use case yet.
Persistent element blur survives reloads, but screen privacy extensions already exist.
Knowledge compounds across videos instead of re-searching from scratch every query.
Social media recompression testing is clever, but steganography tools like Steghide already exist.
Regex-based PII masking for ChatGPT leaks, but detection is intentionally basic.