matplotlib中plt用法實例

 

 

import torch
from models.models import Model
import cv2
from PIL import Image
import numpy as np

from matplotlib.animation import FFMpegWriter
import time
import matplotlib.pyplot as plt


from torchvision.transforms import functional


exp_name = './xxxx_results'
dataRoot = 'xxxx.mp4'
model_path = './checkpoint_best.pth'


def pre_image(image):
    image = Image.fromarray(cv2.cvtColor(image,cv2.COLOR_BGR2RGB))
    input_image = image.copy()
    # image.show()
    height, width = image.size[1], image.size[0]
    height = round(height / 16) * 16
    width = round(width / 16) * 16
    image = image.resize((width, height), Image.BILINEAR)

    image = functional.to_tensor(image)
    image = functional.normalize(image, [0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
    return input_image,torch.unsqueeze(image,0)


if __name__ == '__main__':

    device = torch.device('cuda:0')

    # load model
    model=Model()
    checkpoint = torch.load(model_path)
    model.load_state_dict(checkpoint['model'])

    model.cuda()
    model.eval()

    # input video
    video = cv2.VideoCapture(dataRoot)
    fps = video.get(cv2.CAP_PROP_FPS)
    print(fps)
    frameCount = video.get(cv2.CAP_PROP_FRAME_COUNT)
    print(frameCount)
    size = (int(video.get(cv2.CAP_PROP_FRAME_WIDTH)), int(video.get(cv2.CAP_PROP_FRAME_HEIGHT)))


    # metadata = dict(title='Video Test', artist='Matplotlib', comment='Movie support!')
    # writer = FFMpegWriter(fps=25, metadata=metadata)

        # videoWriter = cv2.VideoWriter('trans.mp4', cv2.VideoWriter_fourcc(*'MP4V'), fps, size)
    success, frame = video.read()
    index = 1

    figure = plt.figure()
    while success:
        # time1=time.time()
        src_image,frame = pre_image(frame)
        images = frame.to(device)

        # time1 = time.time()


        # ground truth
        # gt_path = dataRoot + '/den/' + filename_no_ext + '.csv'

        # predict
        dense_map,atten_map = model(images)
        # test = time.time() - time1

        dense_map = dense_map.cpu().data.numpy()[0,0,:,:]
        # test=time.time()-time1

        dense_pred_count = np.sum(dense_map)
        dense_map = dense_map/np.max(dense_map+1e-20)

        # cv2.imshow("image", dense_map)
        # cv2.waitKey(0)


        plt.subplot(121)
        plt.imshow(src_image)
        # plt.title('original image')
        plt.axis('off')

        plt.subplot(122)
        plt.imshow(dense_map)
        # plt.title('dense map')
        plt.text(25, 25, 'pred crowd count:%.4f ' % dense_pred_count, fontdict={'size': 10, 'color': 'red'})
        plt.axis('off')

        plt.tight_layout(pad=0.3, w_pad=0, h_pad=1)

        # anni=animation.FuncAnimation(fig, animate, init_func=init,frames=200, interval=20, blit=True)
        # anim.save('sin.gif', fps=75, writer='imagemagick')
        plt.savefig(exp_name + '/'+ str('%05d' % index) + '_' + str(int(dense_pred_count)) + '.png', bbox_inches='tight', pad_inches=0, dpi=150)

        # plt.show()
        plt.clf()

        success, frame = video.read()
        index += 1

    video.release()