paddle尝试PP-Vehicle-属性识别

PP-Vehicle-属性识别

windows11环境，conda虚拟环境中运行。
官网地址：PP-Vehicle

1. 首先安装环境->

   • conda create -n paddle python=3.7
   • conda activate paddle
     两种方式，一种是缺啥装即：
   • pip install paddlepaddle-gpu
   • pip install pyyaml
   • pip install opencv-python
   • pip install scipy
   • pip install matplotlib
   • pip install scikit-learn
   • pip install pandas
   • pip install tqdm
   • pip install shapely
   • pip install requests
   • pip install lab

   另外一种是：pip install -r requirements.txt

2. 下载PaddleDetection代码，网址为PaddleDetection

3. 进入到文件夹内，准备测试的视频cd C:\Users\10106\Desktop\PaddleDetection-release-2.6，随便下载一个汽车的视频，放到了demo/test.mp4

4. 下载模型，放到pretrained/vehicle_attribute_model下，predtrain是自己创建的。其他模型最好也都下载好，解压一下

5. 编辑配置文件deploy/pipeline/config/infer_cfg_ppvehicle.yml中VEHICLE_ATTR是用于属性识别的，将其中的model_dir替换为下载的模型。

6. 修改配置文件中deploy/pipeline/config/infer_cfg_ppvehicle.yml 的VEHICLE_ATTR项的enable: True，以启用该功能。

7. 运行代码python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --video_file=demo/test.mp4 --device=gpu，出现了个以下的问题。

8. 测试单张图片python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --device=gpu --image_file=demo/car.jpg

9. 测试文件夹中的图片python deploy/pipeline/pipeline.py --config deploy/pipeline/config/infer_cfg_ppvehicle.yml --device=gpu --image_dir=demo

问题解决

1. 第一次检测视频运行卡在了这里，单张图片可以。将paddle换成gpu版本后，提示Unable to use JDE/FairMOT/ByteTrace提示要安装lab。主要原因为环境安装时出现问题。最好使用pip install -r requirements.txt进行安装，要不然会缺包。
   

PP-Human-行人属性识别

1.使用与上面相同的环境
2.下载对应的模型
3.运行单张图片python deploy/pipeline/pipeline.py --config deploy/pipeline/config/examples/infer_cfg_human_attr.yml --device=gpu --image_file=demo/hrnet_demo.jpg
4. 属性介绍

• 性别：男、女
• 年龄：小于18、18-60、大于60
• 朝向：朝前、朝后、侧面
• 配饰：眼镜、帽子、无
• 正面持物：是、否
• 包：双肩包、单肩包、手提包
• 上衣风格：带条纹、带logo、带格子、拼接风格
• 下装风格：带条纹、带图案
• 短袖上衣：是、否
• 长袖上衣：是、否
• 长外套：是、否
• 长裤：是、否
• 短裤：是、否
• 短裙&裙子：是、否
• 穿靴：是、否
  可在该路径下查看标签内容PaddleDetection-release-2.6\deploy\pipeline\pphuman\attr_infer.py
  5.写了个函数把里面部分转化为中文，并且保存为txt

    def save_txt(self,result,files):
        # - Gender
        # - Age: Less than 18; 18-60; Over 60
        # - Orientation: Front; Back; Side
        # - Accessories: Glasses; Hat; None
        # - HoldObjectsInFront: Yes; No
        # - Bag: BackPack; ShoulderBag; HandBag
        # - TopStyle: UpperStride; UpperLogo; UpperPlaid; UpperSplice
        # - BottomStyle: LowerStripe; LowerPattern
        # - ShortSleeve: Yes; No
        # - LongSleeve: Yes; No
        # - LongCoat: Yes; No
        # - Trousers: Yes; No
        # - Shorts: Yes; No
        # - Skirt&Dress: Yes; No
        # - Boots: Yes; No
        txt = []
        attr_map = {'Male':'男','Female':'女',"Glasses: True":"眼镜","AgeLess18":"小于18岁","Age18-60":"18到60岁之间","AgeOver60":"大于60岁",
                    "Hat: True":"帽子","Backpack":"背包","ShoulderBag":"肩包","HandBag":"手提包","Upper: LongSleeve":"长袖上衣","Upper: ShortSleeve":"短袖上衣",
                    "LongCoat":"长外套","Trousers":"长裤","Shorts":"短裤","Skirt&Dress":"短裙","Boots":"靴子"}
        human_attr_res = result.get('attr')
        print(human_attr_res['output'])
        for human_attr in human_attr_res['output']:
            for i in range(len(human_attr)):
                if human_attr[i] in attr_map.keys():
                    # human_attr[i]  = attr_map[human_attr[i]]
                    txt.append(attr_map[human_attr[i]])
                elif "Upper:" in human_attr[i] or "Lower:" in human_attr[i]:
                    temps = human_attr[i].split(" ")
                    for temp in temps:
                        if temp in attr_map.keys():
                            txt.append(attr_map[temp])
            files[0] = files[0].replace("\\","/")
            txt_name = (files[0].split("/")[-1].split(".")[-2] + '.txt') if "." in files[0] else (files[0] + '.txt')
            with open(txt_name,"w") as file:
                # 将字典转换为字符串，以空格隔开键和值
                dict_str = ' '.join(set(txt))
                file.write(dict_str)