From Turbulence to Tranquility: Crowd as Smooth as Water with Ai
Problem Statement 1 - Weave AI Magic with Groq
CrowdFlow addresses the critical need for real-time crowd density monitoring in public spaces, festivals, religious gatherings, and events. It assists the authorities/organizers/safety personnel in proactively identifying high-risk zones to prevent stampedes and overcrowding.
Brief real-world use case: Imagine a festival or a piligrimage (Mahakumbh or Hajj) with thousands of attendees — CrowdFlow visually monitors the scene, detects density in zones, and offers live insights + spoken alerts to improve crowd management.
R3GE
- Affaan Jaweed (https://www.linkedin.com/in/affaanjaweed/)
- Farhan Khan (https://www.linkedin.com/in/farhan-khan-668439300/)
- Mohammed Abrar (https://www.linkedin.com/in/mohammedabrar934/)
- Syed Aamir (https://www.linkedin.com/in/syed-aamir-8b5591300/)
Why we chose this problem:
Crowd-related disasters are very catastrophic and are uninevitable BUT can be prevented with timely data. We wanted to blend AI vision, real-time feedback, and interactivity to help reduce risks and assist authorities with actionable insights.
- Real-time processing of crowd video streams
- Accurate person detection and zone heatmapping
- Communication via both chat and speech
- Dynamic zone risk assessment
- Used CSRNet instead of YOLO for accurate data
- Integrated GROQ LLM for contextual zone suggestions
- Built custom zone-wise heatmap and TTS flow
- Shifted from basic overlays to intelligent AI commentary
- Frontend: HTML,CSS,JAVASCRIPT
- Backend: Flask(Python
- Database: None (Optionally MongoDB)
- APIs: Groq,OpenAI,PlayAI
- Hosting: Render
- [✅] Groq: How you used Groq
- Monad: Your blockchain implementation
- Fluvio: Real-time data handling
- Base: AgentKit / OnchainKit / Smart Wallet usage
- Screenpipe: Screen-based analytics or workflows
- Stellar: Payments, identity, or token usage (Mark with ✅ if completed)
- ✅ Real-time crowd detection using CSRNet
- ✅ Accurate Crowd Data with count and occupancy % for better visualization
- ✅ Zone-wise heatmap overlays for density visualization
- ✅ Risk analysis with dynamic AI suggestions via LLM
- ✅ Voice output using Text-to-Speech for safety alerts
- Demo Video Link: https://www.youtube.com/watch?v=KU4XNdZjL14
- Pitch Deck / PPT Link: https://docs.google.com/presentation/d/1PWzBTzLtm5iHC6zk_-QWTDmxw5Rs8NNP/edit?usp=sharing&ouid=104295308132445088685&rtpof=true&sd=true
- Deploy Link: https://crowdflow-vtia.onrender.com
- [✅] All members of the team completed the mandatory task - Followed at least 2 of our social channels and filled the form (Details in Participant Manual)
- All members of the team completed Bonus Task 1 - Sharing of Badges and filled the form (2 points) (Details in Participant Manual)
- All members of the team completed Bonus Task 2 - Signing up for Sprint.dev and filled the form (3 points) (Details in Participant Manual)
(Mark with ✅ if completed)
- Python
- API Keys: Groq API Key
- .env file setup (if needed)
# Clone the repo
git clone https://github.com/NoobieDYG/R3GE-CrowdFlow
# Install dependencies
pip -r 'requirements.txt'
# Start development server
python bacakend\app.pyMake sure if the weights are downloaded if not then manually download from the google drive link provided and drag it to vision_model\weights directory GDrive Link : https://drive.google.com/drive/folders/1gvjLW4kKJlqorUgS3Hgf_nKrNiWZBnke?usp=drive_link
- 📈 Add multiple camera feeds / drone integration
- 🛡️ Implement user roles + admin dashboards
- 🌐 Multilingual support for global accessibility
- 📊 Advanced analytics dashboard with historical trends
- API used : Groq,OpenAi
- Groq Documentation,TensorFlow,Documentation,IJCA Research Paper,Jetir Research Paper
- Acknowledgements
Enjoyed this journey very much. Had a beautiful first experience with my team. Learned a lot about collaborating with teammates and building a wonderful project while being inpressure about the deadline Kudos to the NameSpace community for organising this event Cheers!🍻