BawalVision | Industrial Vision & SCADA HMI

🍱

Food Composition Distribution

Detection frequency across standard meal classes

YOLOv8 INT8 Gate: 35%

📡

No meal records logged yet in active session. Awaiting camera stream and detector ingestion.

🔍

QC Defect Breakdown

Distribution of rejection causes across inspected batches

Total Defects: 0

✓

Zero rejection incidents recorded in the current session. All inspected trays meet chromatic and composition criteria.

🛡️

Kitchen Quality Assurance Protocol

Every meal tray must undergo optical quality inspection, chromatic freshness verification, and portion component quota checks before dispatch.

⚡

Hardware & Edge Pipeline Health

ESP32-S3 sensor node telemetry, camera stream, and inference acceleration

○ DISCONNECTED

ESP32-S3 Dual-Core Node WAITING_DEVICE

ESP32 Ambient Temp Offline

CPU 1 (Core 0 - WiFi) 0%

CPU 2 (Core 1 - App) 0%

ESP Free Heap Available

Camera Ingestion Stream Standby / Offline

Throughput Frame 0.0 FPS

End-to-End Latency 0.0 ms

Camera Model Standby

Compute Engine & QC Gate Local Multi-Arch SIMD

Parallel Workers 4 vCPU Threads

Host Memory Usage Active

Active Confidence Gate 35% (Range 45–98%)

⚠️

Flagged & Rejected Trays

Recent audit records requiring supervisory attention

✓

No Flagged or Rejected Trays

All processed meal trays in the current database meet optical, freshness, and food safety standards.

AUDIT & COMPLIANCE

Inspection Reports

Rolling 50-slot evidence buffer with annotated bounding-box snapshots. Export individual trays or the whole batch as Markdown.

Annotated evidence

Inspection Audit Stream 0 records 50-slot rolling image buffer

Loading inspection audit records...

HIGH-THROUGHPUT PARALLEL INSPECTION

Batch Inspection

Upload a ZIP archive of food tray photos. The C++ backend extracts every image, runs YOLOv8s INT8 inference across OpenMP worker threads, and compiles a multi-page LaTeX master report per batch.

IMG_BATCH

OpenMP + ARM NEON

AWS Lambda Mobile Engine: Serverless C++ Runtime (ARM64 · YOLOv8s INT8)

Uploading ZIP

Processing OpenMP Multi-Core

Compiling LaTeX Master PDF

Live ESP32 Inspection View

Standby

📷

ESP32-S3 Live Stream Awaiting Signal

Streaming securely over Cloudflare Camera Relay (WebSocket WSS). Waiting for ESP32 to publish frames.

🍱

Detected Items

0 Items Active

🍽️ Awaiting Food Tray

Place meal tray in camera view for real-time AI food detection & quality inspection.

🍱

Detected Food Items

Dynamic YOLO Object Detection & Freshness Analysis

#	Detected Item	CV Match	ML Spoilage	Decision
No food items detected in camera view.

LINE AUDIT & EVENT LOG

[3:16:09 PM] [INFO] BawalVision UI initialized. Awaiting live telemetry & camera ingress from backend...

📊

QC Inspection Yield

Production Statistics

Awaiting Scan

0% Standby

✓

Accepted 0

Rejected 0

🥗

Food Quality & Freshness

Thermal Stability & Shelf Life

Awaiting Food Scan

Critical (0%) Warning (50%) Optimal (100%)

Quality Score 0 / 100

T-Safe Remaining 8.5 hours

🌡️

Thermal & Hardware Telemetry

ESP-IDF Device Disconnected

Offline

🌡️

-- °C Offline

⚡

-- % Offline

ESP32-S3 Temperature Unavailable (Disconnected)

Core CPU Utilization Unavailable (Disconnected)

⚡

Manual Controls

Actuator Triggers

Armed

📦

Batch Summary

Lot #2026-A

Total Inspected 0 units

Pass Rate 0.0%

☁️

Database Sync

24ms AWS RDS

Local Buffer 0 queued

PostgreSQL RDS Offline

telemetry & cpu utilization

ESP32-S3 Temp CPU 1 (Core 0 - WiFi) CPU 2 (Core 1 - App)

🌡️ Temperature Monitor 15°C – 65°C

Temp

45°C 35°C 25°C 15°C

📡 ESP32-S3 Offline

⚡ Dual CPU Utilization (Core 0: WiFi / Core 1: App) 0% – 100%

CPU %

100% 75% 50% 25% 0%

📡 ESP32-S3 Offline

operational shift timeline

Avg. ESP32 Temp

Offline

Avg. CPU Load

Offline

Device

Disconnected

FMW

Offline

CPU 1 (Core 0)

Offline

CPU 2 (Core 1)

Offline

RSSI

Offline

Heap Free

Offline

Internal RAM

Offline

PSRAM

Offline

Result Log 0 records

ID	Timestamp	Food Composition	Temperature	Status
Loading inspection archives from PostgreSQL...

MONITORING

Grafana Dashboard

Embedded analytics workspace for throughput, spoilage trends, and inference telemetry served from http://localhost:3002.

Open Grafana in New Tab

AI Model Real-Time Monitor

YOLOv8 & CV Engine

AI Ingestion Throughput

24.8 FPS

Crow HTTP + POSIX SHM

YOLOv8s INT8 Latency

14.2 ms

SIMD AVX2 OpenMP Cores

Fleet Acceptance Yield

96.8% PASS

2,360 total meals audited

Spoilage Defect Rate

3.2% FLAGGED

Thermal & visual inspection

SIMD Freshness Index Distribution

AVX2 Freshness Kernel

0-50%

12 (1.2%)

50-60%

24 (2.4%)

60-70%

48 (4.8%)

70-80%

110 (11%)

80-90%

320 (32%)

90-100%

486 (48.6%)

Edge Sensor & Optical Telemetry

ESP32-S3 IoT Node

Internal Node Temp 25.4 °C Optimal

Luminance (Lux) 480 lux Standard

Color Temp (CCT) 5200 K Daylight

Safe Consumption (t_safe) 8.5 hours Fresh

Indonesian Food Dataset Class Telemetry

10 Target Classes

Food Class	Samples	Avg Confidence	Freshness Score	Pass Rate	Quality Status
rice	412	92.4%	98.6%	99.2%	PASS
fried_chicken	384	88.5%	96.2%	97.4%	PASS
tempeh	345	84.1%	95.8%	96.8%	PASS
tofu	290	82.7%	96.4%	98.1%	PASS
boiled_egg	215	87.3%	97.1%	98.5%	PASS
spinach	198	81.2%	94.5%	95.2%	PASS
water_spinach	176	79.8%	93.8%	94.6%	REVIEW
banana	142	94.6%	99.1%	99.4%	PASS
orange	110	89.2%	98.2%	98.9%	PASS
apple	88	91.5%	98.8%	99%	PASS

⚙️

System & Hardware Settings

Sensor Thresholds and Device Nodes

🌐

Gateway & Hardware Connectivity

Fedora Peripheral & Wireless Node Controller

📡

ESP32-S3 Node Management (OTA & NVS Controller)

Directly inspect hardware telemetry, query live device status, and deploy target backend URL into ESP32 NVS

⚡ One-Click Wi-Fi Sync (via USB — COMx/ttyACM*)

Connect ESP32 via USB. Enter target Wi-Fi credentials → click Sync Wi-Fi. The node automatically switches to your network without typing or guessing any IP address!

Target Wi-Fi Network (SSID)

Wi-Fi Password

⚠️ Web Serial is unavailable in this browser. Open via Google Chrome or Edge to pair USB directly.

🖥️

Browser Dashboard Gateway (HMI Backend API)

Specifies where this web dashboard running in your browser fetches AI inference results, batch audits, and inspection reports

For remote access use your laptop LAN IP (e.g. http://<laptop-ip>:8080/api/v1). To run host backend, use just run-backend.

Hardware Serial Ports (`/dev/ttyACM`, `/dev/ttyUSB`)

Direct Linux serial device controller & hardware node inspection

ℹ️

No USB serial ports detected

Plug an ESP32-S3 into the backend host and press Rescan Ports, or use Direct Browser Connect / Web Serial below.

Wi-Fi Networks

Connect ESP32-S3 node

📡

No Wi-Fi networks found

Press Scan Now and verify NetworkManager (nmcli) is running on the backend host.

💡 Select your Wi-Fi network above to provision ESP32-S3, or plug into a detected USB serial port to automatically apply credentials.

🎬

Inspection Source

🎯

Detection Thresholds

Minimum Confidence Gate (35%) Vision classification acceptance threshold (Range: 25% – 98%, Default: 35%)

⚡

Hardware Accelerators

Cloud Inference (ARM64 SVE Acceleration) Unavailable Requires ARM64 cloud hardware instance

Local C++20 Vision Service (OpenMP SIMD) Active on-premise zero-dependency machine vision pipeline (multi-threaded)

Food Composition Distribution

QC Defect Breakdown

Hardware & Edge Pipeline Health

Flagged & Rejected Trays

ESP32-S3 Live Stream Awaiting Signal

Detected Items

Detected Food Items

LINE AUDIT & EVENT LOG

QC Inspection Yield

Food Quality & Freshness

Thermal & Hardware Telemetry

Manual Controls

Batch Summary

Database Sync

SIMD Freshness Index Distribution

Edge Sensor & Optical Telemetry

Indonesian Food Dataset Class Telemetry

System & Hardware Settings

Gateway & Hardware Connectivity

ESP32-S3 Node Management (OTA & NVS Controller)

Browser Dashboard Gateway (HMI Backend API)

Hardware Serial Ports (/dev/ttyACM*, /dev/ttyUSB*)

Wi-Fi Networks

Inspection Source

Detection Thresholds

Hardware Accelerators

Hardware Serial Ports (`/dev/ttyACM`, `/dev/ttyUSB`)