A light weight framework for Object Detection.


Install
pip install eyewitness==1.2.1

Documentation

License: MIT Documentation Status PyPI Version ci pipeline status

EyeWitness

Lightweight Framework for object detection application. wrapper your own detector and add your handler.

system design: https://drive.google.com/file/d/1x_sCFs91swHR1Z3ofS4e2KFz6TK_kcHb/view?usp=sharing

document: https://eyewitness.readthedocs.io/en/latest/index.html

env

support python 2.7, 3.5, 3.6

Installation

pip install eyewitness

manually installation

git clone https://gitlab.com/penolove15/witness.git
cd witness
python setup.py install

Real Detector examples with docker

  • MobileNet(caffe pre-trained) with cv2 repo
  • RefineDet implemented by sfzhang15 with caffe
  • pelee implemented by Robert-JunWang with caffe
  • MTCNN implemented by DuinoDu with caffe
  • RFB-SSD implemented by lzx1413 with pytorch
  • yolo-v3 implemented by qqwweee with keras
  • yolo-v3 implemented by ultralytics with pytorch
  • yolo-v3 implemented by xuwanqi with TensorRT
  • Trident implemented by TuSimple with mxnet
  • CenterNet implemented by xingyizhou with pytorch
  • Arcface implemented by deepinsight with mxnet

please take look at README.md inside docker/ there are examples wrapper a detection model

  • pre-trained weighted
  • naive example for detect a image
  • end2end example with webcam
  • evaluation with dataset

DetectedResults Visualization project: Monitor Reporter

https://github.com/penolove/Flask-Monitor-Reporter

a flask UI used for visualization detection results.

MonitorReporter

[Developer] unit-test

nose2

[Developer] Build Ci Image

cd ci;
docker build -t eyewitnessforci/eyewitness-ci-image:1.2.1 .