❗❗ This repo will no longer be maintained, please visit https://github.com/milvus-io/bootcamp ❗ ❗

Video analysis based on Milvus

Common video analytics systems can automatically identify and track the types of moving targets that appear in the video area, and can monitor multiple targets in the same scene simultaneously. The analyzed video can be used in a wide range of applications, such as advertising recommendations, security and entertainment industries. This project uses YOLOv3 and insightface, combined with frame capturing technique of OpenCV, to recognize logos on objects that appear in the video and track and identify faces.

Environmental preparation

• Milvus 2.0
• pymilvus-orm==2.0.0rc2
• tensorflow==1.14.0
• opencv-python==4.2.0.34

Parameter description

This project contains webservice and webclient. Webservice provides the code for the backend service. Webclient provides the scripts for the frontend interface.

The following describes the important parameters of webservice.

common/config.py

Parameter	Description	Default
MILVUS_HOST	Milvus service IP	127.0.0.1
MILVUS_PORT	Milvus service port	19530
LOGO_DIMENSION	Dimension of logo	2048
FACE_DIMENSION	Dimension of face	512
MYSQL_USER	MySql user name	root
MYSQL_PASSWORD	MySql password	123456
MYSQL_DB	MySql database name	mysql
COCO_MODEL_PATH	Path of YOLOv3 model	None
YOLO_CONFIG_PATH	Path of config file	None
FACE_MODEL_PATH	Path of insightface model	None

Steps

1. Install Milvusv2.0 as described in installation overview.

2. Install MySQL.

3. Pull the source code.

   $ git clone https://github.com/zilliz-bootcamp/video_analysis.git

4. Installation dependencies.

   $ pip install -r requirements.txt

5. Download YOLOv3 model.

   $ cd webservice/yolov3_detector/data
   $ ./prepare_model.sh

6. Start service.

   $ cd ../..
   $ python main.py
   # You are expected to see the following outputs.
   ...
   INFO:     Started server process [71037]
   INFO:     Waiting for application startup.
   INFO:     Application startup complete.
   INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
   ...

   Now, get the API by visiting http://127.0.0.1:8000/docs in your browser.

Results display

Logo recognition

Who is he/she?

Only him/her