hidden text to trigger early load of fonts ПродукцияПродукцияПродукцияПродукция Các sản phẩmCác sản phẩmCác sản phẩmCác sản phẩm المنتجاتالمنتجاتالمنتجاتالمنتجات מוצריםמוצריםמוצריםמוצרים
DevMesh

OpenVINO-YOLOV4

Wu Tianwen

Wu Tianwen

Chengdu, Sichuan

2 0
  • 0 Collaborators

This is implementation of YOLOV4,YOLOV4-relu,YOLOV4-tiny ,YOLOV4-tiny-3l ,Scaled-YOLOv4 in OpenVINO2021.3 ...learn more

Project status: Published/In Market

Artificial Intelligence

Intel Technologies
Intel FPGA, Movidius NCS, OpenVINO

Code Samples [1]

Overview / Usage

Suppported models:

  • YOLOv4 (python C++ demo)
  • YOLOv4-relu (python C++ demo)
  • YOLOv4-tiny (python C++ demo)
  • YOLOv4-tiny-3l (python C++ demo)
  • YOLOv4-CSP (only python demo now)
  • YOLOv4x-mish (only python demo now)

Supported model precision:

  • FP32
  • FP16
  • INT8 Quantization

Optimization strategy:

  • Prune YOLOv4 series model and quickly deploy pruned model in openvino:https://github.com/TNTWEN/Pruned-OpenVINO-YOLO
  • INT8 Quantization using OpenVINO POT. Pruned-YOLOv4 series model+ INT8 Quantization will be very friendly to embedded devices

Methodology / Approach

Based on https://github.com/mystic123/tensorflow-yolo-v3

darknet->tensorflow1.x->OpenVINO

Technologies Used

OpenVINO2020.4 OpenVINO2021.3

OpenVINO POT and Accuracy-checker

ALL intel inference device

Repository

https://github.com/TNTWEN/OpenVINO-YOLOV4

Comments (0)