yongjingim172478

해결됐습니다. 앞의 예제를 통해 digits_model.h5을 직접 만들어 넣으면 안되고 , 원래 들어있던 digits_model.h5를 넣으면 잘돌아갑니다. 구성이 달라보입니다

sample_weight_mode

적정 인식률

답변감사드립니다.&nbsp;

메일문제는 일단 face detection 인식률 높이는 문제를 해결후에 다시 해결해보려 합니다.

재밌어서 모바일버전 강의도 수강해볼까 생각중입니다.

감사합니다.

face_Landmark파일 어디서 찾을수있나요?

<img src="https://cdn.inflearn.com/public/comments/40ba953b-c515-48ef-9095-67e95f9bc19a/20200531_174849.png" title="20200531_174849.png" width="755" height="371" alt="" />
고치고 함수 삭제하고 다시 deploy해서 함수 재생성했습니다.
<img src="https://cdn.inflearn.com/public/comments/c800f75b-58d0-4fa1-8cc0-d5b803e54750/20200531_175503.png" title="20200531_175503.png" width="903" height="638" alt="" />
아래사진은 전체적인 로그값이고,&nbsp; 그 아래는 error 로그값을 클릭해서 캡쳐한것입니다. <img src="https://cdn.inflearn.com/public/comments/391f32d0-75a4-477d-ae77-559f3dea53e3/20200531_175741.png" title="20200531_175741.png" width="1519" height="497" alt="" /><img src="https://cdn.inflearn.com/public/comments/21a3c17f-40e1-45e5-a401-f77d567ba22f/20200531_175802.png" title="20200531_175802.png" width="1519" height="621" alt="" />

함정에 빠졌습니다...

답변감사드립니다.
문제는 정말 사소한거였습니다.
jupyter notebook이 실행되고있을때 Home에 image파일을 추가했었는데
그게 jupyter노트북을 끄고&nbsp; 다시 켜니 이 파일을 이제서야 인식합니다.
home에 파일을 추가하면 다시 노트를 껐다켜야 home에 추가한 파일을 프로그램이 인식하나봅니다.
이런 사소한문제때문에 몇시간을 해매서 그런지 앞으로의 단계들이 멀게만 느껴집니다.
수업 잘듣고있습니다. 감사합니다

Haar GUI강의중 에러

import cv2
import numpy as np
from tkinter import *
from PIL import Image
from PIL import ImageTk
#새로운 팩
from tkinter import filedialog

face_cascade_name = './cv2/data/haarcascade_frontalface_alt.xml'
eyes_cascade_name = './cv2/data/haarcascade_eye_tree_eyeglasses.xml'
file_name = './image/marathon_01.jpg'
title_name = 'Haar cascade object detection'
frame_width = 500
#frame 크기를 바꾸면서 활용
def selectFile():
&nbsp; &nbsp; file_name =&nbsp; filedialog.askopenfilename(initialdir = "./image",title = "Select file",filetypes = (("jpeg files","*.jpg"),("all files","*.*")))
&nbsp; &nbsp; print('File name : ', file_name)
&nbsp; &nbsp; read_image = cv2.imread(file_name)
&nbsp; &nbsp; (height, width) = read_image.shape[:2]
&nbsp; &nbsp; frameSize = int(sizeSpin.get())
&nbsp; &nbsp; ratio = frameSize / width
&nbsp; &nbsp; dimension = (frameSize, int(height * ratio))
&nbsp; &nbsp; read_image = cv2.resize(read_image, dimension, interpolation = cv2.INTER_AREA)
&nbsp; &nbsp; image = cv2.cvtColor(read_image, cv2.COLOR_BGR2RGB)
&nbsp; &nbsp; image = Image.fromarray(image)
&nbsp; &nbsp; imgtk = ImageTk.PhotoImage(image=image)
&nbsp; &nbsp; detectAndDisplay(read_image)
&nbsp; &nbsp;&nbsp;
def detectAndDisplay(frame):
&nbsp; &nbsp; frame_gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
&nbsp; &nbsp; frame_gray = cv2.equalizeHist(frame_gray)
&nbsp; &nbsp; #-- Detect faces
&nbsp; &nbsp; faces = face_cascade.detectMultiScale(frame_gray)
&nbsp; &nbsp; for (x,y,w,h) in faces:
&nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; &nbsp; &nbsp; center = (x + w//2, y + h//2)
&nbsp; &nbsp; &nbsp; &nbsp; frame = cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 4)
&nbsp; &nbsp; &nbsp; &nbsp; faceROI = frame_gray[y:y+h,x:x+w]
&nbsp; &nbsp; &nbsp; &nbsp; #-- In each face, detect eyes
&nbsp; &nbsp; &nbsp; &nbsp; eyes = eyes_cascade.detectMultiScale(faceROI)
&nbsp; &nbsp; &nbsp; &nbsp; for (x2,y2,w2,h2) in eyes:
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; eye_center = (x + x2 + w2//2, y + y2 + h2//2)
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; radius = int(round((w2 + h2)*0.25))
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; frame = cv2.circle(frame, eye_center, radius, (255, 0, 0 ), 4)
&nbsp; &nbsp; #cv2.imshow('Capture - Face detection', frame)
&nbsp; &nbsp; image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
&nbsp; &nbsp; image = Image.fromarray(image)
&nbsp; &nbsp; imgtk = ImageTk.PhotoImage(image=image)
&nbsp; &nbsp; detection.config(image=imgtk)
&nbsp; &nbsp; detection.image = imgtk
&nbsp; &nbsp;&nbsp;
#main
main = Tk()
main.title(title_name)
main.geometry()

read_image = cv2.imread("./image/marathon_01.jpg")
(height, width) = read_image.shape[:2]
ratio = frame_width / width
dimension = (frame_width, int(height * ratio))
read_image = cv2.resize(read_image, dimension, interpolation = cv2.INTER_AREA)

image = cv2.cvtColor(read_image, cv2.COLOR_BGR2RGB)
#bgr을 RGB로 바꿔준다.&nbsp;
image = Image.fromarray(image)
imgtk = ImageTk.PhotoImage(image=image)&nbsp;

face_cascade_name = './cv2/data/haarcascade_frontalface_alt.xml'
eyes_cascade_name = './cv2/data/haarcascade_eye_tree_eyeglasses.xml'

face_cascade = cv2.CascadeClassifier()
eyes_cascade = cv2.CascadeClassifier()

#-- 1. Load the cascades
if not face_cascade.load(cv2.samples.findFile(face_cascade_name)):
&nbsp; &nbsp; print('--(!)Error loading face cascade')
&nbsp; &nbsp; exit(0)
if not eyes_cascade.load(cv2.samples.findFile(eyes_cascade_name)):
&nbsp; &nbsp; print('--(!)Error loading eyes cascade')
&nbsp; &nbsp; exit(0)

label=Label(main, text=title_name)
label.config(font=("Courier", 18))
label.grid(row=0,column=0,columnspan=4)
sizeLabel=Label(main, text='Frame Width : ')&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
sizeLabel.grid(row=1,column=0)
sizeVal&nbsp; = IntVar(value=frame_width)
sizeSpin = Spinbox(main, textvariable=sizeVal,from_=0, to=2000, increment=100, justify=RIGHT)
#정렬은 오른쪽으로
sizeSpin.grid(row=1, column=1)
Button(main,text="File Select", height=2,command=lambda:selectFile()).grid(row=1, column=2, columnspan=2, sticky=(W, E))
detection=Label(main, image=imgtk)
detection.grid(row=2,column=0,columnspan=4)
detectAndDisplay(read_image)

main.mainloop()

정말 텐서플로1 텐서플로2로 구분되어서 강의가 올라와있네요. 답변 감사합니다 . 잘듣고있습니다

yongjingim172478

投稿