/computer-vision

All the Computer Vision Models I've worked on

Primary LanguageJupyter NotebookMIT LicenseMIT

Computer Vision Models

Description:

All models (optimized,pre-built and trained) for my work are listed here

Pre-Built

  • detect_gender - CLI for gender detection on static 2-D image
  • gender_detection_model - Pre-Built Model
  • gender_detect_webccam - CLI for gender detection on video frames
  • smallervggnet - Small VGG Net Model Architecture
  • train - Builder Script
  • Age_Gender_Pre_Built - Based on VGG16 Age Gender Module Built

Facial Recognition

  • data_generator - Script which generates pictures using burst mode through camera
  • Training - Training for similarity using LBPH Recognizer
  • main - Main App which gives similarity metric

Trained

Built on easily availaible datasets and images generated from OpenVINO Toolkit Models

  • build_imfdb - Build Database Script
  • opt_model.h5 - Optimized Model
  • SMALL_VGG_NET - Initial Model Build
  • smallvggnet - optimized Model Build
  • train_model - Training Scipt for Model with multiple outputs
  • init - Model,Graph Initializer
  • convertor - .svg to .png convertor
  • deploy.prototxt - caffemodel specs
  • gen - qrcode generator
  • integrator - Model(s) integrator
  • new_face_detection - Script for GUI Build and interface
  • res10_300x300_ssd - CaffeModel
  • new_model - Model with OpenVINO Trainers
  • xception - Xception Net with IMFDB Trainers
  • Age_Gender_Continous - Age Gender Module (Densenet121) with Continous Age
  • Age_Gender_Categorical - Age Gender Module (Densenet121) with Softmax Ages
  • Age_Gender_One_Hot - Age Gender Module (Densenet121) with One Hot Ages

Human Detector

Built on imageai library with RetinaNet,YoloV3

  • hum_detector - Human Detector on pictures
  • hum_detector_pic_resnet50 - Same as above but with Model Loader Changed
  • hum_detector_self - Detector on vid frames