基于pytorch预训练模型使用Faster RCNN调用摄像头进行目标检测无敌详细！简单！超少代码！

Posted 2022-12-06 猫头丁

基于pytorch预训练模型使用Faster RCNN调用摄像头进行目标检测【无敌详细！简单！超少代码！】

详细完整项目链接：https://download.csdn.net/download/weixin_46570668/86954697?spm=1001.2014.3001.5503

使用 Pytorch 自带的预训练模型 fasterrcnn_resnet50_fpn，该模型是在COCO 数据集上进行预训练。
COCO 数据集是一个大型的、丰富的用于物体检测、分割的数据集，提供的图片类别有80 类，有超过 33 万张图片，其中 20 万张图片有标注，整个数据集中个体的数目超过 150万个。COCO2017 数据共 80 个类别，类别的 id 号不是连续的，最大为 90。

记载预训练模型代码如下：

import torch
import torchvision
model = torchvision.models.detection.fasterrcnn_resnet50_fpn(
    pretrained=True, progress=True, num_classes=91, pretrained_backbone=True)

使用OpenCV调用摄像头，代码如下：

import cv2
capture=cv2.VideoCapture(0)

获得标签及对应类别（这个文件好像可以去COCO还是Pytorch的什么网站上下载，我有点记不清了，不过我也把它和代码一起打包上传资源）

classes_path = 'classes.txt'
classes_path = os.path.expanduser(classes_path)
with open(classes_path) as f:
    class_names = f.readlines()
class_names = [c.strip() for c in class_names]
num_classes = len(class_names)

进行目标检测，并画图函数

def detect_image(image):
    start_time = time.time()
    image_shape = np.array(np.shape(image)[0:2])
    old_width = image_shape[1]
    old_height = image_shape[0]
    old_image = copy.deepcopy(image)
    width, height = get_new_img_size(old_width, old_height)
    image = image.resize([width, height])
    photo = np.array(image, dtype=np.float32) / 255
    photo = np.transpose(photo, (2, 0, 1))
    with torch.no_grad():

        model.eval()
        # Using RGB conversion to convert image pixel-wise into a numpy array
        np_sample_image = np.array(image.convert("RGB"))

        # Converting the numpy array to a tensor
        transformed_img = torchvision.transforms.transforms.ToTensor()(
            torchvision.transforms.ToPILImage()(np_sample_image))
        result = model([transformed_img])

        bbox = []
        label = []
        conf = []
        for i in range(len(result[0]['scores'])):
            if result[0]['scores'][i] > confidence:
                bbox.append(result[0]['boxes'][i].tolist())
                label.append(result[0]['labels'][i].tolist())
                conf.append(result[0]['scores'][i].tolist())
        bbox = np.array(bbox)
        label = np.array(label)
        conf = np.array(conf)
        bbox[:, 0::2] = (bbox[:, 0::2]) / width * old_width
        bbox[:, 1::2] = (bbox[:, 1::2]) / height * old_height
        bbox = np.array(bbox, np.int32)
    image = old_image
    thickness = (np.shape(old_image)[0] + np.shape(old_image)[1]) // old_width * 2
    font = ImageFont.truetype(font='simhei.ttf',
                                  size=np.floor(3e-2 * np.shape(image)[1] + 0.5).astype('int32'))

    for i, c in enumerate(label):
        predicted_class = class_names[int(c)]
        score = conf[i]

        left, top, right, bottom = bbox[i]
        top = top - 5
        left = left - 5
        bottom = bottom + 5
        right = right + 5

        top = max(0, np.floor(top + 0.5).astype('int32'))
        left = max(0, np.floor(left + 0.5).astype('int32'))
        bottom = min(np.shape(image)[0], np.floor(bottom + 0.5).astype('int32'))
        right = min(np.shape(image)[1], np.floor(right + 0.5).astype('int32'))

        # 画框框
        label = ' :.2f'.format(predicted_class, score)
        draw = ImageDraw.Draw(image)
        label_size = draw.textsize(label, font)
        label = label.encode('utf-8')
        print(label)

        if top - label_size[1] >= 0:
            text_origin = np.array([left, top - label_size[1]])
        else:
            text_origin = np.array([left, top + 1])

        for i in range(thickness):
            draw.rectangle(
                    [left + i, top + i, right - i, bottom - i],
                    outline=colors[int(c)])
        draw.rectangle(
                [tuple(text_origin), tuple(text_origin + label_size)],
                fill=colors[int(c)])
        draw.text(text_origin, str(label, 'UTF-8'), fill=(0, 0, 0), font=font)
        del draw

    print("time:", time.time() - start_time)
    return image

依次读入帧，转变格式，并送入模型进行检测

t1 = time.time()
    # 读取某一帧
    ref,frame=capture.read()
    # 格式转变，BGRtoRGB
    frame = cv2.cvtColor(frame,cv2.COLOR_BGR2RGB)
    # 转变成Image
    frame = Image.fromarray(np.uint8(frame))
    frame = np.array(detect_image(frame))

    frame = cv2.cvtColor(frame,cv2.COLOR_RGB2BGR)
    fps  = ( fps + (1./(time.time()-t1)) ) / 2
    print("fps= %.2f"%(fps))

    cv2.imshow("video",frame)
    c= cv2.waitKey(30) & 0xff
    if c==27:
        capture.release()
        break

完整代码如下！赶紧试试看！

import numpy as np
import time
import torch
import torchvision
import os
import cv2
import copy
import colorsys
from PIL import Image, ImageFont, ImageDraw
# 调用摄像头
capture=cv2.VideoCapture(0)

fps = 0.0

model = torchvision.models.detection.fasterrcnn_resnet50_fpn(
    pretrained=True, progress=True, num_classes=91, pretrained_backbone=True)

def get_new_img_size(width, height, img_min_side=600):
    if width <= height:
        f = float(img_min_side) / width
        resized_height = int(f * height)
        resized_width = int(img_min_side)
    else:
        f = float(img_min_side) / height
        resized_width = int(f * width)
        resized_height = int(img_min_side)

    return resized_width, resized_height

# classes_path = 'voc_classes.txt'
classes_path = 'classes.txt'
classes_path = os.path.expanduser(classes_path)
with open(classes_path) as f:
    class_names = f.readlines()
class_names = [c.strip() for c in class_names]
num_classes = len(class_names)
mean = torch.Tensor([0,0,0,0]).cuda().repeat(num_classes+1)[None]
std = torch.Tensor([0.1, 0.1, 0.2, 0.2]).cuda().repeat(num_classes+1)[None]
# 画框设置不同的颜色
hsv_tuples = [(x / len(class_names), 1., 1.)
              for x in range(len(class_names))]
colors = list(map(lambda x: colorsys.hsv_to_rgb(*x), hsv_tuples))
colors = list(
    map(lambda x: (int(x[0] * 255), int(x[1] * 255), int(x[2] * 255)),
                colors))
#---------------------------------------------------#
#   获得所有的分类
#---------------------------------------------------#

confidence = 0.5
def detect_image(image):
    start_time = time.time()
    image_shape = np.array(np.shape(image)[0:2])
    old_width = image_shape[1]
    old_height = image_shape[0]
    old_image = copy.deepcopy(image)
    width, height = get_new_img_size(old_width, old_height)
    image = image.resize([width, height])
    photo = np.array(image, dtype=np.float32) / 255
    photo = np.transpose(photo, (2, 0, 1))
    with torch.no_grad():

        model.eval()
        # Using RGB conversion to convert image pixel-wise into a numpy array
        np_sample_image = np.array(image.convert("RGB"))

        # Converting the numpy array to a tensor
        transformed_img = torchvision.transforms.transforms.ToTensor()(
            torchvision.transforms.ToPILImage()(np_sample_image))
        result = model([transformed_img])

        bbox = []
        label = []
        conf = []
        for i in range(len(result[0]['scores'])):
            if result[0]['scores'][i] > confidence:
                bbox.append(result[0]['boxes'][i].tolist())
                label.append(result[0]['labels'][i].tolist())
                conf.append(result[0]['scores'][i].tolist())
        bbox = np.array(bbox)
        label = np.array(label)
        conf = np.array(conf)
        bbox[:, 0::2] = (bbox[:, 0::2]) / width * old_width
        bbox[:, 1::2] = (bbox[:, 1::2]) / height * old_height
        bbox = np.array(bbox, np.int32)
    image = old_image
    thickness = (np.shape(old_image)[0] + np.shape(old_image)[1]) // old_width * 2
    font = ImageFont.truetype(font='simhei.ttf',
                                  size=np.floor(3e-2 * np.shape(image)[1] + 0.5).astype('int32'))

    for i, c in enumerate(label):
        predicted_class = class_names[int(c)]
        score = conf[i]

        left, top, right, bottom = bbox[i]
        top = top - 5
        left = left - 5
        bottom = bottom + 5
        right = right + 5

        top = max(0, np.floor(top + 0.5).astype('int32'))
        left = max(0, np.floor(left + 0.5).astype('int32'))
        bottom = min(np.shape(image)[0], np.floor(bottom + 0.5).astype('int32'))
        right = min(np.shape(image)[1], np.floor(right + 0.5).astype('int32'))

        # 画框框
        label = ' :.2f'.format(predicted_class, score)
        draw = ImageDraw.Draw(image)
        label_size = draw.textsize(label, font)
        label = label.encode('utf-8')
        print(label)

        if top - label_size[1] >= 0:
            text_origin = np.array([left, top - label_size[1]])
        else:
            text_origin = np.array([left, top + 1<
以上是关于基于pytorch预训练模型使用Faster RCNN调用摄像头进行目标检测无敌详细！简单！超少代码！的主要内容，如果未能解决你的问题，请参考以下文章
 
 pytorch Faster-RCNN 的验证损失
 使用ResNet101作为预训练模型训练Faster-RCNN-TensorFlow-Python3-master
 使用ResNet101作为预训练模型训练Faster-RCNN-TensorFlow-Python3-master
 基于预训练模型的Unet超级简单懒人版Pytorch版
 基于预训练模型的Unet超级简单懒人版Pytorch版
 使用 Pytorch 保存在 Faster RCNN（COCO 数据集）上训练的最佳模型，避免“过度拟合”