Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

I’m Too Lazy to Google Search an Image and Paste it Back in Word, So I Built EZSearch

Description and use

EZSearch automatically finds and inserts images into documents, cutting search time from an hour to ten minutes by using local Llama 3.2 3B and CLIP ViT‑B/32 models.

Problem

Users waste time searching for suitable images to paste into documents, often spending up to an hour to find non-blurry, non-watermarked photos that match a sentence.

Outcome

EZSearch reduced image search and paste time from ~1 hour to ~10 minutes, with the first search taking ~13 s and subsequent searches ~1 s thanks to caching. The tool operates entirely offline for query building and scoring, sending only a short query over the network.

Workflow

  • Select a sentence in any document or browser.
  • Press Ctrl+Shift+X to trigger EZSearch.
  • A popup appears with up to eight matching images.
  • Click an image to paste it into the document; Ctrl‑click opens the source; right‑click copies attribution.
  • The sentence is sent to a local LLM (Llama 3.2 3B) to build a query; images are fetched; CLIP scores them; the best image is pasted.
  • Cached results speed up repeat searches.

Evidence

The sentence is read from the clipboard and handed to the local Ollama server, so it never leaves the machine.

Source

~13 s for the first search of a new subject on this laptop, with no GPU and no API.

Source

I was able to reduce my time to just 10 minutes, taking roughly 30‑45 seconds to show suggested images.

Source

The sentence is read from the clipboard and handed to a local Ollama server, so it never leaves the machine.

Source

  • Record date: 2026-10-24T21:11:32.009Z