NOBAR AYAM — Detection by Case (nobar-ayam)

On-demand visual review for suspect chicken batches. Upload a recorded batch video, run inference with the same YOLO pose model as production (per unit), and download an annotated review video.

Does not replace or interfere with the real-time jetson-counter service (port 5000).

Brand	NOBAR AYAM — Nonton Bareng, Review Bareng!
Default port	5050
LAN URL (this site)	http://192.168.192.96:5050
Install path	/opt/nobar-ayam

Reference: System_Plan_Detection_by_Case.md

Features

• Choose unit: Salatiga or Cicalengka (different models/config)
• Upload MP4 / AVI / MOV (max 500MB default)
• Single-job queue (one video at a time)
• Progress + ETA in browser
• Annotated output: skeleton, boxes, IDs, counting line, HUD
• Temp files in /tmp/nobar-ayam — cleaned after download

Deploy di 192.168.192.96 (port 5050)

Jalankan di host Jetson/LAN 192.168.192.96. Port 5050 dipilih agar tidak bentrok dengan dashboard counting di 5000.

# 1. Salin kode
sudo mkdir -p /opt/nobar-ayam
sudo cp -r nobar-ayam/* /opt/nobar-ayam/
sudo chown -R jetson:jetson /opt/nobar-ayam
cd /opt/nobar-ayam

# 2. Config
cp config.env.example .env
nano .env   # sesuaikan CASE_SALATIGA_MODEL / CASE_CICALENGKA_MODEL
sed -i 's/\r$//' .env

# 3. Dependency di venv counter (torch/ultralytics sudah ada)
/opt/jetson-counter/venv-counter/bin/pip install flask opencv-python-headless

# 4. Systemd
sudo cp nobar-ayam.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now nobar-ayam

# 5. Verifikasi
sudo systemctl status nobar-ayam --no-pager
curl -s http://127.0.0.1:5050/api/units
# Browser: http://192.168.192.96:5050

Setelah ubah .env:

sudo systemctl restart nobar-ayam

Log:

sudo journalctl -u nobar-ayam -f

Jika firewall aktif:

sudo ufw allow 5050/tcp
sudo ufw reload

Quick start (manual, tanpa systemd)

cd /opt/nobar-ayam
cp -n config.env.example .env
source /opt/jetson-counter/venv-counter/bin/activate
export PYTHONNOUSERSITE=1
python app.py

Open: http://192.168.192.96:5050

Environment variables

Variable	Default	Purpose
CASE_PORT	5050	Flask port
CASE_HOST	0.0.0.0	Bind address (LAN)
CASE_TEMP_DIR	/tmp/nobar-ayam	Temp upload/output
CASE_MAX_UPLOAD_MB	500	Max upload size
CASE_SALATIGA_MODEL	/media/jetson/DATA/yolo11n-salatiga.engine	Salatiga model
CASE_CICALENGKA_MODEL	/media/jetson/DATA/yolo11n-cicalengka.engine	Cicalengka model
CASE_IMGSZ	416	Inference size
CASE_HALF	true	FP16 (TensorRT)
CASE_CONF	0.3	Detection confidence
CASE_DEVICE	0	GPU index

File contoh: config.env.example. App juga membaca .env di folder yang sama saat start.

API

Endpoint	Method	Description
/	GET	Upload UI
/api/units	GET	List units
/api/jobs	POST	Upload unit + video
/api/jobs/<id>	GET	Job status + progress
/api/jobs/<id>/download	GET	Download annotated MP4
/api/jobs/<id>/cleanup	POST	Manual temp cleanup

Project layout

File	Role
app.py	Flask routes + .env loader
case_processor.py	Video inference + overlay
case_queue.py	Single-job worker queue
case_config.py	Per-unit settings
case_batch.py	In-memory batch counting
overlay.py	Drawing helpers
templates/index.html	Web UI (branded)
static/logo.png	Brand logo
static/app.js / app.css	Frontend
nobar-ayam.service	systemd unit
config.env.example	Env template

Notes

• Run on-demand only — avoid heavy jobs while real-time counting is under stress.
• Model per unit must match the site where the video was recorded.
• Queue is in-memory; restarting the app clears pending jobs.
• Port map: counting dashboard 5000, NOBAR AYAM 5050.

Related docs

• jetson-counter/README.md — production counter
• jetson-counter/PANDUAN-GO-LIVE.md — magang onboarding