Development of video surveillance system software

Budget: 1000 USD Deadline: 30 days

Hello, Oleksandr!
Thank you for the detailed project brief — it is well thought out, and the task is ambitious and extremely interesting. I specialize in developing computer vision systems, integrating with Telegram, video processing, and working with AI/ML. I am ready to implement a system that will operate 24/7, efficiently utilize resources, and be conveniently managed through Telegram 📲

🚀 Architecture
🔹 Stream separation:
Sub-stream (720p) — primary detection on TPU (YOLO/MobileNet SSD).
Main stream (FullHD) — continuous recording on HDD without re-encoding (mp4), synchronization with sound (optional).
🔹 TPU Coral — accelerates object detection (people, cars, animals) with minimal load on CPU/GPU.

🎯 Intelligent analysis
✅ Triggers initiate analysis of the Main stream (± a few seconds).
✅ CPU+GPU:
Face recognition (matching with the “Known” database).
Optionally: tracking, car color/model.
✅ Audio analysis — detection of loud sounds, scream/glass break (optional).

📐 Zones, rules, filters
🔸 Customizable polygon zones.
🔸 Rules engine:
Conditions: time, zone, object, face, sound.
Actions: log, alert, ignore.

📌 Example:
“Perimeter, 23:00-06:00, person, Not Known → Telegram”

🗃️ Storage and Database
🗄 PostgreSQL:
Cameras, zones, rules, “Known” face vectors, event log.
API for adding “Known” faces.

💽 HDD: structured video archive.
🔁 Auto-cleaning/rotation of the archive.
🤖 Telegram bot
📲 Alerts with text, screenshot, buttons:
[Known] – add to database,
[Alert] – escalation,
[Ignore] – filtering.

💬 Commands:
/status – system status,
/ get_event – details,
/ get_log – recent events.

🔐 Authorization via Telegram ID.
🌐 Web interface (optional)
Settings for cameras, zones, “Known” faces, rules.
Viewing events, logs, archives, live video.
🧰 Technologies
Python – the foundation of the system.
CV/ML: OpenCV, TensorFlow Lite, PyTorch.
RTSP: GStreamer, FFmpeg.
Bot: python-telegram-bot.
API: FastAPI.
Containers: Docker for deployment.
Monitoring: systemd + auto-start.

📦 Result
🔹 Ready Docker image or installation instructions
🔹 Source code with comments
🔹 Demonstration on your equipment
🔹 Complete documentation
🔹 Further support (by agreement)

💼 Why me?
✅ Experience in video analytics systems, ML, CV, Telegram API
✅ Focus on performance, security, and flexibility
✅ Ability to build scalable architecture
✅ Promptness and openness to dialogue

Respectfully, Semen!
🔧 Python Dev / CV Engineer Expand

Budget: 1000 USD Deadline: 30 days

Hello!
I am an AI engineer, with a main specialization in computer vision.
I have experience in developing my own neural networks.
For your task, I can use YOLO.
I implement object detection, including face recognition of people.

Write to me personally, and we can discuss more details.

Budget: 25 USD Deadline: 1 day

Hello, I have experience working with IP cameras... the project is quite extensive... and I would like to learn more about the team) my main profile is computer vision (python) with 4+ years of experience, I am waiting for you in private messages to discuss the details!

Budget: 1250 USD Deadline: 45 days

Worked on similar tasks: video processing, detection on TPU, face recognition on CPU/GPU, integration with a Telegram bot (python-telegram-bot), logic of triggers and rules, storage in PostgreSQL. Well acquainted with load balancing. Used Docker for deployment. The technical specifications look familiar – I will do it without problems.
The proposed conditions are approximate and require adjustments after the final acceptance of all aspects of the project.