跳到正文
DEV Community· InApp·· 7 天前AI 評分39

Shorts 過濾器測試 9,000 段影片,現行模型攔截率達 99.98%

99.98% blocked: I ran 9,000 Shorts through my filter so the old model could fail in peace

AI 導讀

作者以 9,000 段 Shorts 測試過濾器,現行模型攔截率為 99.98%;舊版模型則放過 8,667 段,漏攔率為 96.3%。過濾器先抽取每段影片的 8 個畫面並去重,再進行文字稿與視覺嵌入處理;若省略去重,成本約為現行方式的 12 倍,攔截率相同。

正文

Cover image for 99.98% blocked: I ran 9,000 Shorts through my filter so the old model could fail in peace

InApp

I built this filter because my own Shorts feed had become unusable — scammy re-uploads, AI slop loops, and clickbait compilations stacked ten deep. I wanted numbers, not vibes, so I pulled 9,000 Shorts and scored them against two classifiers: the current one and the previous revision I still had deployed.

The old model let 8,667 of the 9,000 through. That's a 96.3% miss rate, and it explains every "why is my feed like this" moment I've had this year. The current model blocks 99.98%.

One concrete detail of how it works: I don't classify the video file directly. I sample 8 frames per video, hash them, and dedupe near-identical frames across the whole corpus first — that alone kills most mass-reupload spam before any model sees it. Survivors go to a transcript-plus-visual embedding pass, and anything under the confidence threshold gets blocked, not demoted. Demotion just moves the junk into the "maybe" rail where humans still see it.

What it costs the operator: roughly 40 GPU-minutes per 1,000 videos on the frame-sampling pass, plus storage for the hashes. If I skipped the dedupe step and sent everything straight to the model, that bill would be about 12x higher for the exact same block rate. The dedupe layer is the whole economics of the thing.

I recorded the full walkthrough with the per-class breakdowns and the failure cases of the old model: https://www.youtube.com/shorts/oZmarp6b8Qw

來源:DEV Community · dev.to