Видео

While reshaping a model, how do we determine which index holds the originality?

😅

can we do this using pytorch?

concise, unlike many other youtube tutorials

How to use these features for object detection or othe processing.please share the next video link

that was a great presentation

What if i have 2 varibles input and 1 variable output, how i can make a prediction of these constanta?

thank you so much, the presentation is so helpful and clear, it helps me a lot.

Hey.... This video was very helpful. Can you please make a video on DataSets. Like describing all the methods of using DataSet such as from online source, from your storage, and especially the ones in csv format..

Great video for making image patches

This code was helpful for me. Dhanywaad Devi Ji

Nice Presentation!

Thank you so much. This helped me a lot😊

Hi i have an open question : if i augment my image data with this library, is there a way to also apply them to the labels ? For example if i do a rotatation or a flip , i also have to modify the bounding boxes. Can this library do this?

Please mam show that how to save each image patch in a folder.

can you share your notebook ?

Thanks ! for the details explanation. Its really useful

import os import cv2 import numpy as np import skvideo skvideo.setFFmpegPath('anaconda3/envs/test/bin/') import skvideo.io print ("hi") # caminho para o vídeo de entrada path2video = (r"C:\Users\alif\Desktop\Optical_Flow\VIDEOS/Tudo_bem_3.mp4") outputPath = (r"C:\Users\alif\Desktop\Optical_Flow\OptFlowMaps") # read the video to create frames-generator videogen = skvideo.io.vread(path2video) print(f'shape of video {path2video} = {videogen.shape}') # reading first frame as first and previous frames # New image (array) filled with zeros in same dimensions as input frame (3 chanels) frame_ind = 0 hsv = np.zeros_like(videogen[frame_ind]) # conver input frame from RGB to gray prevF = cv2.cvtColor(videogen[frame_ind], cv2.COLOR_RGB2GRAY) print("Reading frames...") # range com a largura de videogen for index in range(len(videogen)): # printando a informação de qual frame está sendo processado por iteração print(f"{index}/{len(videogen)}") # conver input frame from RGB to gray nextF = cv2.cvtColor(videogen[index], cv2.COLOR_RGB2GRAY) # maintain similar size of next anf previous frames; shape[1] é a width, e shape[2] é a high dim = (prevF.shape[1], prevF.shape[0]) nextF = cv2.resize(nextF, dim) #computing optical flow map by Farneback method. Returns 2D vector of size input frame # cv.calcOpticalFlowFarneback(prev, next, flow, pyr_scale, levels, winsize, iterations, poly_n, poly_sigma, flags ) flow = cv2.calcOpticalFlowFarneback(prevF, # prev nextF, # next None, # flow 0.5, # pyr_sacel 3, # levels 15, # winsize 3, # iterations 5, # poly_n 1.2, # poly_sigma 0) # flags # Computando a magnitude e o ângulo dos vetores 2D mag, ang = cv2.cartToPolar(flow[..., 0], flow[..., 1]) # Setting hue image de acordo com o a direção do fluxo óptico hsv[..., 0] = (((ang * 180) / (np.pi)) / 2) # setting hue image intensities de acordo com a magnitude do fluxo óptico normalizado hsv[..., 2] = cv2.normalize(mag, None, 0, 255, cv2.NORM_MINMAX) # Convertendo hue image from HSV to BGR format bgr = cv2.cvtColor(hsv, cv2.COLOR_HSV2BGR) # fluxo óptico é salvo com o nome do frame dest = outputPath + str(index) + ".png" cv2.imwrite(dest, bgr) # Fluxo óptico se torna o frame anterior e loop se resume no novo nextF prevF = nextF

very useful video..if you could also show how to save each image patch in a folder that would be great

Excellent video! It helped me a lot :3

It helps me a lot in my project

how to combine or concatenate these features ??

excelente video, muchas gracias, pude resolver varios problemas de "RIFE"

i know the pixel width in meters. now i have a sequence of point-symmetrical images. i want to measure a speed field. #skyimager

Nice work. Would you please share your source code?

I need code

Will you able to upload code ?

I need code

hi kindly share ur email, i am training model VGG16 but getting low validation accuracy

Great work