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

# Summary The author experimented with using computer vision and cursor automation to automate Instagram engagement by taking screenshots, detecting UI elements like heart icons through visual matching rather than DOM selectors, and clicking them programmatically. The approach works around Instagram's obfuscated HTML by treating the platform like any visual interface a human can see, using landmark detection (like the triple-dots menu and action bar) to locate where hearts appear rather than searching the entire screen. The author warns that while technically feasible, Instagram's anti-abuse systems will detect and ban accounts using such automation, making it impractical as an actual growth strategy.