How it Works

If you are a manufacturer, look for AI algorithm running on your AI cameras.

If you are a developer of a startup company, look for AI cameras for your AI weights

If you are an installer or a system integrator, look for AI cameras for problem solving



Manufacturer

Developer

  • Train your YOLO AI model
  • Convert your YOLO model
  • Deploy your YOLO model


    •

End User

  • Buy LILIN Z7/P7 series cameras
  • Select proper AI models
  • Configure events and meta-data
  • Install the camera
    •

What to Do

Visit LILIN open GtiHub for more model here...

|Benefits

Object tracking & PTZ tracking algorithms

Color detections

SmartEvent for IO triggering

AES256 weight protection

Object left, object removal, tripwire, density detection, counting

|Why Edge AI

The Era of Edge Computing

|Cybersecurity

AES256 firmware encryptions for AI firmware and AI model

Model encryption and protection for intellective property

AI firmware encryption and protection for cyberecurity

Developer

If you are a developer, look for self-trained YOLO camera.

|Step #1 Label the pictures

Create object names

Label your pictures using the name

|Step #2 Training YOLO

Download this CFG file to train your YOLO dataset


|Step #3 Convert your YOLO Model

This AI cloud is for the AI developer to drop files (1) YOLO CFG files, (2) YOLO AI model and (3) Your .names file here for the conversion of the trained YOLO AI weights (models) running on LILIN Aida cameras. At the time of YOLO weight conversion, every classification needs 10 labled pictures with .names files and the JPEG files for calibrating of AI pruning & sparsification, all the pictures and labled files will be removed from LILIN AI Cloud aftering weight conversion. LILIN will not store and reuse these files.


Drag & drop YOLO CFG, weights, and object list here.

Open File Explorer
File List
  • No files uploaded.
  • Your YOLO CFG file (*.cfg) is required.
  • Your YOLO object list file (*.names) is required.
  • Your YOLO weights file (*.weights) is required.
  • Your partial training images (*.jpg) are required.
  • Your logo(32*32) (*.png) is optional.

|Step #4 packaging yolo ai firmware

AI firmware LPKG packing information

All files will be removed on LILIN AI Cloud at 1:00 AM, Taiwan time


Ready to convert!


AI weight conversion status
  • Loading plugin






|Step #5 Download to AI cameras

Manufacturer


If you are a manufacturer, look for AI engines run on your cameras.

|How to Apply

Sign the NDA agreement in between your company and LILIN

Permission to use logo of your company to this forum

Purchase LILIN AI camera for testing purpose

Release LILIN Aida plug-in container SDK for running YOLO weight

Sell your Ambarella CV products with YOLO weight built-in

# Model Object Behavior Network
1 Traffic Bicycle, car, motorbike, airplane, bus, train, truck,
boat, tuk-tuk, pickup truck, fire engine,
container, VAN, SUV, police car, ambulance		 Traffic flow counting
Left turn flow counting
Right turn flow counting
Wrong way counting
Queuing detection
YOLO v3 416
2 Fisheye Traffic Bicycle, car, motorbike, airplane, bus, train, truck,
boat, tuk-tuk, pickup truck, fire engine,
container, VAN, SUV, police car, ambulance		 Car park counting YOLO v3 416
3 Car Make Recognition More than 85 brands of car Counting YOLO v3 tiny 416
4 Human Detection Person, mask, no mask, improper mask wearing Person, mask, no mask, improper mask wearing YOLO v3 tiny 416
5 Construction Site Safety Helmet, no helmet, not proper helmet wearing,
safety vest detection		 Helmet, no helmet, not proper helmet wearing,
safety vest detection		 YOLO v3 tiny 416
6 Train solution Pantograph, person Pantograph detection YOLO v3 tiny 416
7 Fire and smoke Fire, smoke Detection zone for fire and smoke YOLO v3 tiny 416
8 LPR Australia
Austria, Belgium, Italy, Croatia, France, Turkey, Germany, UK
Indonesia,
Nepal,
India,
Japan,
Saudi Arabia, UAE, Iran, Oman, Dubai,
Malaysia, Hong-Kong, Macau, Singapore
Taiwan,
USA: 50 states, Guam, and Puerto Rico, Vietnam
Thailand
Sri Lanka (coming soon)
Support state and country names for Indonesia, Nepal, Bangladesh, India, Japan, USA and Thailand YOLO v3 416

System Installer / System Integrator


If you are an installer or system integrator, looking for AI cameras for problem solving

Contact LILIN distributors or branches for purchasing P7/Z7 series cameras

Edge AI Camera Performance (INT32 -> FX8)


# Feature Model Performance
1 License plate recognition (LPR) 2 AI models 6 FPS recognitions
2 License plate recognition (LPR) + traffic object 3 AI models 5 FPS recognitions
3 License plate recognition (LPR) + traffic object + car make 4 AI models 3 FPS recognitions
4 Object recognition 1 AI models 9~12 FPS recognitions

|The secondary development

Visit LILIN Open Source GitHub for development

Visit Python code for RTSP video and AI bounding boxes displaying on video for edge AI camera


|The "Terms of Service" for LILIN AI Cloud

We will not claim ownership of any of your AI weights. The intent of LILIN AI Cloud is to convert your AI weights for running on LILIN IP camera.

We will not use and save your AI weights for marketing or promotional advertising activities.

We will not save your AI weights on LILIN AI Cloud.