Building a real-time crowd density analyzer based of off Kishore et al work 🫨 A system to prevent crowd disasters at events: >detects people >tracks movement >maps high-density zones in real-time. Going from theory to deployment, one phase at a time

Built a CLI tool that crawls your site and finds broken links before your users do 🔗📷 -> works on localhost before you push -> fast, scriptable link validation without opening a browser -> drop it in CI/CD to block deploys with dead links -> exports JSON reports -> respects robots.txt Run without installing: npx @initysl/checker yoursite.com Install globally: npm install -g @initysl/checker checker yoursite.com npm: npmjs.com/package/@initysl/c… GitHub: github.com/initysl/checker

Love building interactive frontend 👨🏻‍🍳🔥⛷️🏻 > made with react and framer motion 🍻

Built a CLI tool that crawls your site and finds broken links before your users do 🔗⁉️ -> works on localhost before you push -> fast, scriptable link validation without opening a browser -> drop it in CI/CD to block deploys with dead links -> exports JSON reports -> respects robots.txt Run without installing npx @initysl/checker yoursite.com Install globally npm install -g @initysl/checker checker yoursite.com Built with Node.js from scratch - crawler, parser, reporter, the works. npm: npmjs.com/package/@initysl/c… GitHub: github.com/initysl/checker

Replicated Mariia Petrovych 'Where To Go' Cards Animation' swipeable card stack in Next.js Framer Motion 🃏 The entire animation runs on one number ->`offset` (distance from the top card): - scale, rotateZ, translateY all derive from offset -> depth, tilt, and peek in 3 lines - active card uses useMotionValue for live drag -> background cards use static computed values - useTransform maps vertical drag -> rightward drift for that natural swipe-away feel - on release, if y > 180px the array rotates -> first card moves to last. Infinite loop, zero resets - initial === animate on every card ->recycled cards snap into place, no awkward re-entrance UI Credit: Mariia Petrovych:dribbble.com/shots/25468979-…

Lamine Yamal 🇵🇸 💙 ❤️ 💛

I’m currently leveraging the InsightFace library to build a streamlined facial surveillance system!🥂 The pipeline is quite simple as at now: >load a target image -> detected once >image got embedded -> stored in memory >video -> processed frame-by-frame -> each frame's faces get detected -> each detected face gets embedded -> compare frame embedding with stored target embedding -> draw a bounding box on detected target's face -> discard frame embedding after comparison Eager to share more details as it progress! 🥂🫡

Mujeres 10 minutos después de sacar el carnet de conducir VS Hombres 10 minutos después...

Introducing WireBird. Autonomous drones that perch, recharge, and relaunch from powerlines. We’re building a retrofit system that lets drones use existing electrical infrastructure as their charging network. Security sites could have drones perched like birds on wires, always charged and ready to respond. Utility corridors, highways, farms, and remote infrastructure could be monitored by drones that recharge from the lines they follow. Schools, hospitals, bases, and critical sites could deploy low-cost protection domes anywhere power infrastructure already exists. We just finished prototype V2. If you know customers, companies, or use cases across utilities, infrastructure, security, agriculture, defense, or public safety we’d love to talk

Thought i'd share this, may come in handy! 🙂 When managing LLM limits, you can use the 3/4 rule to estimate your remaining output: since one token is approximately 4 characters or 3/4 of a word (or 1,000 tokens = 750 words), you can quickly gauge how much text the model will generate before hitting a rate limit or truncation point. >formula: remaining Tokens * 0.75 = available words (outputs) >3/4 = 0.75 >example: If the API indicates 400 tokens remaining, you can expect roughly 300 words of output. 🥂

Crowd density project update 🚨📹, and of course, a bird's-eye and high-angle view is best for this system. Details shown in the demo output explained: >green zones: safe crowd density ✅ >yellow zones: caution (getting packed) ⚠️ >red zones: critical (danger threshold) 🚨 >direction arrows: where people are moving >magenta boxes: bottlenecks detected (packed barely moving) >orange warnings: reverse flow (panic indicator) Testing high-angle surveillance shows promising results, though the bird's-eye view still needs some tuning. The system tracks individual IDs, calculates max density per zone, and flags danger in real-time.

Yeah, It’s true that perspective, occlusion, and weather are tough bottlenecks. However, modern pros counter these by using camera calibration to "fix" tricky angles and density maps to count crowds when people are bunched together. By adding a loop of active learning, you can keep the model sharp even when the environment changes. Ultimately, the secret to success is treating deployment as a continuous learning process rather than a one-time handoff.

Just had a session with Supervision and it’s a game-changer for CV! 🚀🎯 The workflow is so much cleaner: >detection >tracking >tracing

My goal is to sell one canvas this month🙏🏼 please share as much as you can 😫

Hey, my name is Oumaima If the algorithm showed you my art, i appreciate any support i get🙌🏻

Alhamdulillah, SeekDeen.com is live. I built it as a simple Islamic learning companion with Quran, duas, Hanafi fiqh, hadith, seerah, the 99 Names, memorization, and guided Ask support. The goal is to make learning Deen easier, calmer, and more accessible.

CV 👀 📸