마이펫: 반려견의 사료 및 물 먹는 횟수 확인(Clova Custom Extension 스킬 구현)

 

• Clova Custom Extension 인터렉션 모델 및 발화문(Slot, Intent) 등록 및 테스트

• 전체 아키텍처 구성(AWS 및 딥러닝 서버 구축 및 서비스 설계) 

• IoT Cam 구축(motioneyeos, Raspberry PI, mjpeg 스트리밍 사용) 

• 컴퓨터 비전(Object Detection, Yolov4)을 통한 반려동물 행동 확인

 

• IoT Cam 

  - Raspberry Pi 3B+

  - motioneyeos

 

• Client

  - Clova iOS App

 

• Server

  - AWS EC2, ALB(for SSL), Route 53

  - Deep learning Server(Nvidia Geforce RTX 2080)

  - Docker 기반 서비스 구현(flask API framework, gunicorn, redis)

 

• 반려견이 밥이나 물을 먹었는지 Clova 앱 음성인식을 통해 확인 가능함.

 

• Git 

  - IoT Cam: 참조 https://github.com/ccrisan/motioneyeos 

  - web-ui(Clova Interface구축): https://github.com/asyncbridge/web-ui 

  - pet-detector(Object Detection): https://github.com/asyncbridge/pet-detector 

  - redis-store: https://github.com/asyncbridge/redis-store 

• Homepage: https://ai.bakevision.com/task04 

• 데모 영상: https://youtu.be/ufnZmMCBYIg