DVA_LAB

Drone Video Analysis(DVA) project for helping MARC supported by Kakao Impact, MODULABS

1. Repository Structure

DVA_LAB
│
├── backend
├── models
│ ├── BEV
│ ├── bytetrack
│ ├── efficientAD
│ ├── refiner
│ └── sahi_detection
├── docs
├── frontend
└── README.md

2. backend

This directory contains the backend logic of the DVA system, including APIs and services for data and model operations.

api/routers

• data_router.py: API routes for handling data-related requests.
• model_router.py: API routes for model-related operations.

services

• data_service.py: Services for data manipulation and processing.
• model_service.py: Services for interacting with the analysis models.

utils

• remove_glare: 빛반사 제거 (to be updated)
• log_sync: This module is for reading metadata and srt file from .MP4, .SRT. (to be updated)
• visualizing: This file is for visualizing the result of detection and tracking. it shows a kind of dashboard at top-left side of frame. especially has the vessel violated?, distance between vessel and dolphin, speed of vessel and so on. (to be updated)

3. models

This directory contains different model directories, each operating as a microservice as part of a Microservices Architecture (MSA) to ensure modularity and avoid dependency conflicts. Each model serves a specific analysis purpose:

• BEV: Implements a microservice for Bird's Eye View transformations and analysis, providing a top-down perspective of video data.
• bytetrack: A microservice for tracking objects in video data, allowing for continuous object identification across frames.
• efficientAD: An efficient anomaly detection microservice model designed to identify unusual patterns or anomalies in video data.
• refiner: A microservice that refines detection or tracking results, improving the accuracy and reliability of the analysis.
• sahi_detection: A dedicated microservice for detection operations, capable of identifying a wide range of objects in various conditions.

Each microservice is self-contained, with its own dedicated environment and dependencies, ensuring that updates or changes to one service do not impact the others, in line with the principles of MSA.

4. frontend

A React application for interactive video analysis.

Contributors ✨

Thanks go to these wonderful people (emoji key):

SeoWoo Han 🚧 💻📖 👀🚇	JunHyung Kang 💻🚇📖👀	HanEol Lee 💻 📖 👀	JiYoung Lee 💻🚇📖 👀	MyeongHun Lim 💻📖 👀	WooHyun Jeon 💻 📖 👀	JongSu Choi 💻 📖 👀
DongWoo Kim 💻 📖 👀	HyeMin Park 💻📖👀	Sunghun Yang 💻	Dongki Chung 💻

This project follows the all-contributors specification. Contributions of any kind are welcome!