All Projects → xialuxi → yolov5_face_landmark

xialuxi / yolov5_face_landmark

Licence: GPL-3.0 license
基于yolov5的人脸检测,带关键点检测

Programming Languages

python
139335 projects - #7 most used programming language
Dockerfile
14818 projects

Projects that are alternatives of or similar to yolov5 face landmark

deep utils
An open-source toolkit which is full of handy functions, including the most used models and utilities for deep-learning practitioners!
Stars: ✭ 73 (-54.09%)
Mutual labels:  yolov5
ros-yolo-sort
YOLO v3, v4, v5, v6, v7 + SORT tracking + ROS platform. Supporting: YOLO with Darknet, OpenCV(DNN), OpenVINO, TensorRT(tkDNN). SORT supports python(original) and C++. (Not Deep SORT)
Stars: ✭ 162 (+1.89%)
Mutual labels:  yolov5
yolov5-dnn-cpp-python
用opencv的dnn模块做yolov5目标检测,包含C++和Python两个版本的程序
Stars: ✭ 326 (+105.03%)
Mutual labels:  yolov5
YOLOv5-Qt
本项目为基于yolov5的GUI目标识别程序,支持选择要使用的权重文件,设置是否使用GPU、置信度阈值等参数。
Stars: ✭ 61 (-61.64%)
Mutual labels:  yolov5
food-detection-yolov5
🍔🍟🍗 Food analysis baseline with Theseus. Integrate object detection, image classification and multi-class semantic segmentation. 🍞🍖🍕
Stars: ✭ 68 (-57.23%)
Mutual labels:  yolov5
yolov5 tensorrt int8
TensorRT int8 量化部署 yolov5s 模型,实测3.3ms一帧!
Stars: ✭ 112 (-29.56%)
Mutual labels:  yolov5
epic-awesome-gamer
🍷 Gracefully claim weekly free games and monthly content from Epic Store.
Stars: ✭ 600 (+277.36%)
Mutual labels:  yolov5
yolov5-opencv-cpp-python
Example of using ultralytics YOLO V5 with OpenCV 4.5.4, C++ and Python
Stars: ✭ 122 (-23.27%)
Mutual labels:  yolov5
yolov5 ROS
ROS使用YOLOv5 run YOLOv5 in ROS
Stars: ✭ 77 (-51.57%)
Mutual labels:  yolov5
yolov5 deepsort tensorrt
This repo uses YOLOv5 and DeepSORT to implement object tracking algorithm. Also using TensorRTX to transform model to engine, and deploying all code on the NVIDIA Xavier with TensorRT further.
Stars: ✭ 117 (-26.42%)
Mutual labels:  yolov5
YOLO-Streaming
Push-pull streaming and Web display of YOLO series
Stars: ✭ 56 (-64.78%)
Mutual labels:  yolov5
pylabel
Python library for computer vision labeling tasks. The core functionality is to translate bounding box annotations between different formats-for example, from coco to yolo.
Stars: ✭ 171 (+7.55%)
Mutual labels:  yolov5
dockerfile-yolov5-jetson
Dockerfile for yolov5 inference on NVIDIA Jetson
Stars: ✭ 30 (-81.13%)
Mutual labels:  yolov5
Vehicle-Detection
Vehicle Detection Using Deep Learning and YOLO Algorithm
Stars: ✭ 96 (-39.62%)
Mutual labels:  yolov5
JSON2YOLO
Convert JSON annotations into YOLO format.
Stars: ✭ 222 (+39.62%)
Mutual labels:  yolov5
reflective-clothes-detect-yolov5
reflective-clothes-detect-dataset、helemet detection yolov5、工作服(反光衣)检测数据集、安全帽检测、施工人员穿戴检测
Stars: ✭ 295 (+85.53%)
Mutual labels:  yolov5
yolov5 cpp openvino
用c++实现了yolov5使用openvino的部署
Stars: ✭ 222 (+39.62%)
Mutual labels:  yolov5
YOLOv4MLNet
Use the YOLO v4 and v5 (ONNX) models for object detection in C# using ML.Net
Stars: ✭ 61 (-61.64%)
Mutual labels:  yolov5
make-your-yolov5 dataset
💥Make your yolov5 dataset by using labelimg.I hope my work can help you make your yolov5 datasets more quickly.
Stars: ✭ 60 (-62.26%)
Mutual labels:  yolov5
dofbot-jetson nano
Yahboom DOFBOT AI Vision Robotic Arm with ROS for Jetson NANO 4GB B01
Stars: ✭ 24 (-84.91%)
Mutual labels:  yolov5

yolov5_face_landmark

基于yolov5的人脸检测,带关键点检测

代码说明:

  • 1,在yolov5的检测基础上,加上关键点回归分支,请先下载yolov5的工程:https://github.com/ultralytics/yolov5
  • 2,detect_one.py是单张图片的测试代码, 基于部分wideface训练的模型,稍后在百度云公开。

主要修改代码部分:

  • (1)hyp.scatch.yaml中增加关键点loss的超参数(landmark: 0.5)
  • (2)yolo.py中增加了关键点回归的计算
  • (3)face_datasets.py为人脸数据的读取方式,准备数据的格式参考yolov5的格式,在后面增加关键点的坐标(归一化)
  • (4) loss.py中增加关键点回归的loss计算
  • (5) 链接: https://pan.baidu.com/s/1zjPIF2NZ9CGtB2iUCox6hw 密码: j83n
  • (6) 效果图 : 效果图

关于口罩人脸的问题:

  • 1,增加口罩人脸这个类别,建议不要直接在检测分支中增加类别。
  • 2,应该在关键点分支额外增加一个属性分支,接一个二分类,判断有没有戴口罩。
  • 3,这样可以减少口罩人脸的误检问题

关于关键点的问题:

  • 1,建议可以替换成wingloss训练,可以优化关键点的精准度。
  • 2,可以解决关键点的离群点问题
  • 3,wideface之中有不少特别小的人脸,如果不处理会对精度有一定的影响。
  • 完整的代码可以参考: https://github.com/deepcam-cn/yolov5-face
Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].