ad-detection

哔哩哔哩浏览器插件：基于大语言模型，对B站视频中的植入广告进行检测。一键跳过视频中的植入/口播广告。

Telegram 群组自助解封机器人 | 基于 Cloudflare Workers + D1 | 多群管理 / 广告检测 / 黑名单导出 / GKY联查 / 314条离线测试 | Fork of CF-Workers-TGUnbanBot【项目取用cmliu/TGUnbanBot】

It detects ads on websites, displays their number and helps with cookie management.

Fights ad-related search results and scam/advertised sites.

Cross-platform commercial detection and automatic TV muting: logo, loudness, black-frame, silence, and perceptual-fingerprint detectors driving IR, CEC, RS-232, or network mute control.

Multimodal Segmentation of Long-Form Online Video into Content and Non-Content

A multimodal machine learning framework utilizing CLIP, BLIP-2, YOLOv8, and EasyOCR to analyze and detect authentic vs. deceptive visual advertisements.

A fork of last30days-skill — this fork rebuilds the Xiaohongshu module into an anti–soft-ad detector: it tells genuine reviews from paid ads, then hands you a briefing, not a scorecard.

TG_Sentinel_bot is a versatile AI-powered Telegram tool that filters messages in channels of any type based on customizable prompts. Detect ads, spam, or any content you define, transferring only safe messages while keeping your channels clean, smart, and fully under your control.

Self-hosted podcast pipeline — downloads YouTube channels + podcast feeds, strips ads with SponsorBlock + LLM (Whisper + Claude), serves clean RSS over Tailscale

Research project for CLEF 2025 Touché Lab on “Advertisements in Retrieval-Augmented Generation.” Covers Subtask 1 (ad generation via RAG pipeline with FAISS, CrossEncoder, Qwen) and Subtask 2 (ad detection using fine-tuned DeBERTa and RoBERTa models). Ranked 2nd and 4th internationally, presented at CLEF 2025 in Madrid.

An intelligent, privacy-focused computer vision system designed to detect and flag advertisement elements in real-time using lightweight edge models