How to automate Instagram engagements with computer vision (and get banned)

He taught a bot to tap Instagram hearts, and commenters say the real click was self-destruct

TLDR: The article shows how to make a computer visually find Instagram’s heart button and click it like a human, dodging fragile website code but risking an account ban. Commenters were split between calling social platforms overly hostile and mocking the whole thing as spammer nonsense that will end badly.

A developer basically posted a how-to guide for making a computer act like a very determined Instagram user: take a screenshot, visually spot the little heart button, move the mouse there, click, repeat — and, as the title gleefully admits, probably get banned. The big idea is simple enough for non-coders: if websites keep changing their behind-the-scenes code, just ignore the code and copy what a human sees on screen instead. Clever? Yes. Suspicious? Oh, absolutely.

And the comments wasted no time turning this into a mini morality play. One camp was instantly annoyed that social platforms have become so aggressive that even making a normal account can feel like applying for airport security clearance. Another camp was far less sympathetic, basically saying: what is this, a spammer starter pack? One of the sharpest jabs called it "a spam post here about how to spam Instagram," which is the kind of line that lands like a folding chair in online discourse.

Then came the nerdy side-quest drama: a commenter got hung up on the article claiming a “typical” screen has over 7 million pixels, leading to a very relatable "wait, are we all secretly on fancy 4K screens now?" moment. Meanwhile, veterans of automation showed up like battle-scarred survivors to say they’ve tried this game before and it ends the same way: banned, annoyed, and posting manually anyway. In short, the community verdict was a mix of grim laughter, spam panic, and pixel-count nitpicking — which, honestly, is the internet at its finest.

Key Points

  • The article says Instagram’s changing HTML structure makes DOM-selector-based automation unreliable.
  • It proposes bypassing page code entirely by using screenshots, visual detection, and cursor automation to interact with UI elements.
  • The article states that hardcoded click coordinates fail because Instagram post layouts shift based on captions, locations, and carousels.
  • It identifies the triple-dots menu and the action bar as stable visual landmarks for narrowing the search area for heart icons.
  • It describes using template matching and a sliding-window search inside the cropped region to reduce false positives when detecting hearts.

Hottest takes

"instantly banned and you have to go through a whole review and approval process" — latchkey
"always get banned at the end. Not worth it" — FitchApps
"A spam post here about how to spam Instagram" — add-sub-mul-div
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.