paddle-CV 人体检测、关键点检测、行为识别、摔倒检测、打架识别、进出管理

最新推荐文章于 2024-02-20 23:50:10 发布

原创最新推荐文章于 2024-02-20 23:50:10 发布 · 1.4k 阅读 ·

本内容遵循CC 4.0 BY-SA版权协议

代码可运行

PP-Human是PaddleDetection框架下用于高频场景的核心工具，支持图片、视频等多种输入，应用于智慧交通和智慧城市等。它提供异常行为识别、人体属性分析、人流计数和跨镜头跟踪等功能，可在T4服务器上实现TensorRT加速的实时处理。

PaddleDetection深入探索核心行业的高频场景，提供了行人场景的开箱即用分析工具，支持图片/单镜头视频/多镜头视频/在线视频流多种输入方式，广泛应用于智慧交通、智慧城市、工业巡检等领域。支持服务器端部署及TensorRT加速，T4服务器上可达到实时！
PP-Human支持四大产业级功能：五大异常行为识别、26种人体属性分析、实时人流计数、跨镜头（ReID）跟踪。

依赖

tqdm
typeguard
visualdl>=2.2.0
opencv-python
PyYAML
shapely
scipy
terminaltables
Cython
pycocotools
setuptools

# for vehicleplate
pyclipper

# for mot
lap
motmetrics
sklearn
filterpy

代码

import gradio as gr
import base64
from io import BytesIO
from PIL import Image
import numpy as np
import os
from pipeline.pipeline import pp_humanv2


# UGC: Define the inference fn() for your models
def model_inference(input_date, avtivity_list):
    
    if isinstance(input_date, str):
        if  os.path.splitext(input_date)[-1] not in ['.avi','.mp4']:
            return None
            
    if 'do_entrance_counting'in avtivity_list or 'draw_center_traj' in avtivity_list:
        if 'MOT' not in avtivity_list:
            avtivity_list.append('MOT')
    
    result = pp_humanv2(input_date, avtivity_list)

    return result


def clear_all():
    return None, None, None


with gr.Blocks() as demo:
    gr.Markdown("PP-Human Pipeline")

    with gr.Tabs():

        with gr.TabItem("image"):

            img_in = gr.Image(value="https://paddledet.bj.bcebos.com/modelcenter/images/PP-Human/human_attr.jpg",label="Input")
            img_out = gr.Image(label="Output")

            img_avtivity_list = gr.CheckboxGroup(["ATTR"])
            img_button1 = gr.Button("Submit")
            img_button2 = gr.Button("Clear")

        with gr.TabItem("video"):

            video_in = gr.Video(value="https://paddledet.bj.bcebos.com/modelcenter/images/PP-Human/human_attr.mp4",label="Input only support .mp4 or .avi")
            video_out = gr.Video(label="Output")

            video_avtivity_list = gr.CheckboxGroup(["MOT","ATTR","VIDEO_ACTION","SKELETON_ACTION","ID_BASED_DETACTION","ID_BASED_CLSACTION","REID",\
                                                    "do_entrance_counting","draw_center_traj"],label="Task Choice (note: only one task should be checked)")
            video_button1 = gr.Button("Submit")
            video_button2 = gr.Button("Clear")

    img_button1.click(
        fn=model_inference,
        inputs=[img_in, img_avtivity_list],
        outputs=img_out)
    img_button2.click(
        fn=clear_all,
        inputs=None,
        outputs=[img_in, img_out, img_avtivity_list])

    video_button1.click(
        fn=model_inference,
        inputs=[video_in, video_avtivity_list],
        outputs=video_out)
    video_button2.click(
        fn=clear_all,
        inputs=None,
        outputs=[video_in, video_out, video_avtivity_list])

demo.launch(share=True, server_port=7015)