def likeBatchImagesByHashtag(self, hashtag: str, next_max_id: str, api, network): img_size = 200 if (next_max_id == None): api.getHashtagFeed(hashtag) else: print("getting images with next_max_id:" + next_max_id) api.getHashtagFeed(hashtag, next_max_id) imgURLs = ImageLoader.getImageURLs(api.LastJson) # get next max ID next_max_id = ImageLoader.getNextMaxID(api.LastJson) img_path = "images/tmp/img.jpg" # calculate like probability per image for url in tqdm(imgURLs): # print(url[0]) try: urllib.request.urlretrieve(url[0], img_path) except: print("loading of image failed") img = dataLoader.load_image(img_path, img_size) # tqdm.write("{}: Calculate prediction...".format(datetime.now())) probability, str_label, label = network.predict(img) # tqdm.write("{0}: Prediction: {1}, Probability: {2:.5f}".format(datetime.now(), str_label, probability)) if str_label == "like": self.like(api, url) time.sleep(2) if cfg.mode == 2: self.likePicturesByUsername(api, url, network) elif cfg.mode == 3: self.likeLikersPictures(api, url, network) return next_max_id
def nextImg(self): self.cur_idx += 1 cur = self.list_idx % self.maxLen if cur not in self.img_dict: print("Warning: Not in dictionary. Returning previous image") worker = ImageLoader.Worker( self.loadImg, self.img_list[self.list_idx], cur) # Any other args, kwargs are passed to the run function self.threadpool.start(worker) self.list_idx += 1 return self.current print("Current: ", self.list_idx, cur, self.img_dict[cur][1]) img = self.img_dict[cur][0] self.current = img if self.list_idx < self.vidLen: # Load new image worker = ImageLoader.Worker( self.loadImg, self.img_list[self.list_idx], cur) # Any other args, kwargs are passed to the run function self.threadpool.start(worker) self.list_idx += 1 else: self.list_idx += 1 return img
def go(): """ Takes a picture from the configured camera and displays the image with recognized faces and labels Parameters: ----------- None Returns: -------- None; shows the image with captioned faces """ img = ImageLoader.get_img_from_camera() dets = ImageLoader.find_faces(img) descs = ImageLoader.find_descriptors(img, dets) compared = ImageCompare.compare_faces(descs, database.database) draw_faces(dets, compared, img)
def _readMask(self): img = ImageLoader.load( self._maskTextureName) #self.maskTexture.image#.convert("RGBA") imgstring = img.convert("RGBA").tostring("raw", "RGBA", 0, 1) buffer = array.array('B', imgstring) self.maskWidth, self.maskHeight = img.size self.mask = buffer
def load(self, begin=None, end=None): if begin == None: start = self.list_idx if self.list_idx < self.maxLen else self.list_idx - self.maxLen load_len = self.maxLen + start if self.maxLen + start < self.vidLen else self.vidLen elif begin != None: start = begin #if end > self.vidLen: # self.list_idx = start + self.maxLen + 1 # return if self.list_idx >= self.vidLen: self.list_idx = start + self.maxLen + 1 return load_len = start + self.maxLen + 1 if start + self.maxLen + 1 < self.vidLen else self.vidLen self.list_idx = start img = Image.open(self.img_list[start]) img = self.prepareImg(img) self.current = img print("Loading...", start, load_len) for i in range(start, load_len): worker = ImageLoader.Worker( self.loadImg, self.img_list[self.list_idx], self.list_idx % self.maxLen ) # Any other args, kwargs are passed to the run function self.threadpool.start(worker) self.list_idx += 1
def add_file(filepath): """ Adds a person to the database given a picture of their face Will ask for their name Parameters ---------- filepath (string): the location of the file that is the picture of the person's face Returns: -------- None """ img = ImageLoader.get_img_from_file(filepath) det = ImageLoader.find_faces(img) descriptor = ImageLoader.find_descriptors(img, det) database.add_image(descriptor)
def main(): argv = sys.argv if len(argv) > 1: if argv[1] == "-youtube": url = input("YoutubeのURLを入力してください。") yt = YouTube(url) name = "download-" + yt.filename yt.set_filename(name) video = yt.get('mp4') video.download("/output/") name = "/output/" + name print(name) else: print("Unknown command") sys.exit() else: name = input("動画パスを入力してください。").replace(" ", "") #元動画の読み込み cap = cv2.VideoCapture(name) if not cap.isOpened(): print("Something wrong...") sys.exit() retval, frame = cap.read() h, w, channels = frame.shape # 出力動画の設定 # 保存ビデオファイルの準備 # Define the codec and create VideoWriter object fourcc = cv2.VideoWriter_fourcc(*'XVID') rec = cv2.VideoWriter('output.avi', fourcc, 20.0, (w, h)) # AVIファイルのフレーム数を取得する frames = int(cap.get(cv2.CAP_PROP_FRAME_COUNT)) while (cap.isOpened()): ret, frame = cap.read() if ret == True: frame = im.find(frame, movie=True) rec.write(frame) # 経過を確認するために100フレームごとに経過を出力 if cap.get(cv2.CAP_PROP_POS_FRAMES) % 100 == 0: date = datetime.datetime.now().strftime("%Y/%m/%d %H:%M:%S") print(date + ' 現在フレーム数:' + str(int(cap.get(cv2.CAP_PROP_POS_FRAMES))) + ' / ' + str(frames), "\r", end="") else: break cap.release() rec.release() cv2.destroyAllWindows()
def allPlot(): # try me ~ 7 seconds per image storage = IL.ImgStorage("../datas/train/") imgNB = storage.size() // 3 fig, axes = plt.subplots(imgNB, 2, figsize=(20, 80)) for i in range(imgNB): print("Computing image ", (i + 1), "out of ", imgNB, "...") img = storage.getImgByIndex(i) res = cleanWingImg(img) axes[i, 0].imshow(img) axes[i, 1].imshow(res) print("Done.") plt.show()
def get_test_data_set(): ''' 获得测试数据集 ''' image_loader = ImageLoader('t10k-images.idx3-ubyte', 10000) label_loader = ImageLoader('t10k-labels.idx1-ubyte', 10000) return image_loader.load(), label_loader.load()
def get_training_data_set(): ''' 获得训练数据集 ''' image_loader = ImageLoader('train-images.idx3-ubyte', 60000) label_loader = ImageLoader('train-labels.idx1-ubyte', 60000) return image_loader.load(), label_loader.load()
def likeLikersPictures(self, api, url, network): likers_count = 1 img_path = "images/tmp/img.jpg" api.getMediaLikers(url[1]) imageLikers = ImageLoader.getImageLikers(api.LastJson) for liker in imageLikers: if likers_count % 7 == 0: return elif not liker[1]: #check if account is private if not liker[0] == 321882169: #check if it is my account self.likeLikersPicture(api, liker[0], network) time.sleep(2) likers_count = likers_count + 1
def rawImgToCSV(imgPath, csvPath): img = ImageLoader.openPathToNpArray(imgPath) print("Cleaning image...") cleanedImg = cleanWingImg(img) print("computing skeleton...") cleanedImg = cleanedImg / 255 skell = skeletonize(cleanedImg) print("keeping centroids...") dotsImg = keepGT4(skell) print("Computing centroids...") centroidList = imageDotsToCentroidList(dotsImg) print("Writing to file...") centroidList = np.flip(centroidList,axis= 1) np.savetxt(csvPath, centroidList, delimiter=",") print("Done.")
def _extractShadowMap(self): #shadowMapTex = TextureManager.GetTexture( self.fsPath()+'shadows.png' ) shadowimage = ImageLoader.load(self.fsPath()+'shadows.png') img = shadowimage.convert("RGBA") #img = shadowMapTex.image.convert("RGBA") imgstring = img.tostring("raw", "RGBA", 0, 1) buffer = array.array('B', imgstring) self.shadowMapWidth, self.shadowMapHeight = img.size self.shadowMap = [1] * (self.shadowMapWidth * self.shadowMapHeight) i1 = 0 i2 = 0 for y in xrange(self.shadowMapHeight): for x in xrange(self.shadowMapWidth): self.shadowMap[i1] = buffer[i2] / 255.0 i1 += 1 i2 += 4
def likeLikersPicture(self, api, userId, network): img_path = "images/tmp/img.jpg" api.getUserFeed(userId) userImagesURLs = ImageLoader.getImageURLsByUsername(api.LastJson) count_likes = 0 for url in tqdm(userImagesURLs): urllib.request.urlretrieve(url[0], img_path) img = dataLoader.load_image(img_path, img_size=200) probability, str_label, label = network.predict(img) if str_label == "like": self.like(api, url) time.sleep(2) count_likes = count_likes + 1 if count_likes >= 2: return
def matcher(fingerfile, fp_database, size, savefolder, ImageLoader, cropping, sectornorm, gabor2d, match): import ImageLoader as IL import cropping as cr import bob import os import sectornorm as sn import gabor2d as gb2 import math import match as mat sp = bob.sp np = bob.ip.numpy #loads data and saves the fingerprint feature information into a database errors = [] fp_number = len(fp_database) f = open(fingerfile, 'r') data_files = f.readlines() data_names = data_files #for imagecount2 in range(0,1): for imagecount2 in range(0, len(data_files)): toLoad = data_files[imagecount2][0:len(data_files[imagecount2]) - 1] print('Loading Image ' + toLoad) print('Image number ' + str(imagecount2 + 1) + ' of ' + str(len(data_files))) #loads selected image into an array and resizes it to be divisible by 8 loadedImage, height, width = IL.loadImage(toLoad) #Crops image around center point #print('Cropping image ' + toLoad) CroppedPrint, errors = cr.crop(height, width, loadedImage, errors, toLoad, size) #normalizes and sectorizes print #print('Sectorizing image ' + toLoad) NormalizedPrint, vector = sn.sector_norms(CroppedPrint, 0, 1, size) #performs gabor analysis and prepares data to be matched. Fingercode are the features and df is the name of the fingerprint in the database fingercode, df = gb2.gabor2d(NormalizedPrint, fp_database, size, imagecount2, data_names, data_files, 1) #matches image against other images. best_match = mat.matcher(savefolder, fp_number, data_files, fp_database, df, fingercode, imagecount2) print('complete')
def loadTextures(): # Configure vars global textureNames pos = 0 tempIDs = [] # For every texture in the list of names, load that file from the textures directory and assign an ID for it for name in textureNames: ID = glGenTextures(1) img_data = ImageLoader.load("Textures/" + name + ".png") tempIDs.append((ID, name)) glBindTexture(GL_TEXTURE_2D, ID) glPixelStorei(GL_UNPACK_ALIGNMENT,1) glTexImage2D( GL_TEXTURE_2D, 0, 3, img_data[0], img_data[1], 0, GL_RGBA, GL_UNSIGNED_BYTE, img_data[2] ) return tempIDs
def templatecreator(fingerfile, savefile, size, ImageLoader, cropping, sectornorm, gabor2d): import ImageLoader as IL import cropping as cr import bob import os import sectornorm as sn import gabor2d as gb2 import pickle sp = bob.sp np = bob.ip.numpy #loads data and saves the fingerprint feature information into a database subtract = 0 errors = [] fp_database = {} f = open(fingerfile, 'r') data_files = f.readlines() data_names = data_files for imagecount in range(0, len(data_names) - subtract): #for imagecount in range(0,1): toLoad = data_files[imagecount][0:len(data_files[imagecount]) - 1] print('Loading Image ' + toLoad) print('Image number ' + str(imagecount + 1) + ' of ' + str(len(data_names) - subtract)) #loads selected image into an array and resizes it to be divisible by 8 loadedImage, height, width = IL.loadImage(toLoad) #Crops image around center point CroppedPrint, errors = cr.crop(height, width, loadedImage, errors, toLoad, size) #normalizes and sectorizes print #print('Sectorizing image ' + toLoad) NormalizedPrint, vector = sn.sector_norms(CroppedPrint, 0, 0, size) #performs gabor analysis and prepares data to be entered into database fp_database, df = gb2.gabor2d(NormalizedPrint, fp_database, size, imagecount, data_names, data_files, 0) # print('Added to Database fingerprint ' + df) pickle.dump(fp_database, open(savefile, 'wb')) return data_files, fp_database
def matcher(fingerfile, fp_database,size, savefolder, ImageLoader, cropping, sectornorm, gabor2d, match): import ImageLoader as IL import cropping as cr import bob import os import sectornorm as sn import gabor2d as gb2 import math import match as mat sp = bob.sp np = bob.ip.numpy #loads data and saves the fingerprint feature information into a database errors = [] fp_number = len(fp_database) f = open(fingerfile, 'r') data_files = f.readlines() data_names = data_files #for imagecount2 in range(0,1): for imagecount2 in range(0,len(data_files)): toLoad = data_files[imagecount2][0:len(data_files[imagecount2])-1] print('Loading Image '+ toLoad) print('Image number ' + str(imagecount2+1) + ' of ' +str(len(data_files))) #loads selected image into an array and resizes it to be divisible by 8 loadedImage,height,width= IL.loadImage(toLoad) #Crops image around center point #print('Cropping image ' + toLoad) CroppedPrint, errors = cr.crop(height,width, loadedImage, errors, toLoad,size) #normalizes and sectorizes print #print('Sectorizing image ' + toLoad) NormalizedPrint, vector = sn.sector_norms(CroppedPrint,0,1,size) #performs gabor analysis and prepares data to be matched. Fingercode are the features and df is the name of the fingerprint in the database fingercode, df = gb2.gabor2d(NormalizedPrint,fp_database,size,imagecount2,data_names,data_files,1) #matches image against other images. best_match = mat.matcher(savefolder,fp_number,data_files, fp_database, df, fingercode,imagecount2) print('complete')
def likePicturesByUsername(self, api, url, network): like_count_by_user = 0 img_path = "images/tmp/img.jpg" api.getUserFeed(url[2]) userImagesURLs = ImageLoader.getImageURLsByUsername(api.LastJson) for url in tqdm(userImagesURLs): urllib.request.urlretrieve(url[0], img_path) img = dataLoader.load_image(img_path, img_size=200) probability, str_label, label = network.predict(img) if str_label == "like": self.like(api, url) like_count_by_user = like_count_by_user + 1 time.sleep(2) if like_count_by_user >= 4: api.follow(url[2]) print("followed user" + str(url[2])) with open('followed_users.txt', 'a') as outfile: outfile.write("\n" + str(url[2])) outfile.close()
def __init__(self, dim, start_url): self.start_url = start_url self.url_frontier = [] self.url_history = [] self.img_urls_history = [] self.image_loader = il.ImageLoader(dim) # header object, necessary to get proper response when requesting urls self.header = { 'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.11 (KHTML, like Gecko) Chrome/23.0.1271.64 Safari/537.11', 'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8', 'Referer': 'https://cssspritegenerator.com', 'Accept-Charset': 'ISO-8859-1,utf-8;q=0.7,*;q=0.3', 'Accept-Encoding': 'none', 'Accept-Language': 'en-US,en;q=0.8', 'Connection': 'keep-alive' }
def LoadTextureAdvanced( filename ): image = ImageLoader.load(filename) ix = image.size[0] iy = image.size[1] image = image.convert("RGBA") img_str = image.tostring("raw", "RGBA", 0, 1) # Create Texture texture = glGenTextures(1) glBindTexture(GL_TEXTURE_2D, texture) # 2d texture (x and y size) glPixelStorei(GL_UNPACK_ALIGNMENT,1) #glTexImage2D(GL_TEXTURE_2D, 0, 4, ix, iy, 0, GL_RGBA, GL_UNSIGNED_BYTE, img_str) gluBuild2DMipmaps(GL_TEXTURE_2D, 4, ix, iy, GL_RGBA, GL_UNSIGNED_BYTE, img_str ) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR) glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_LINEAR) GraphicsCard.setTextureAnisotropy( 8.0 ) glTexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_MODULATE) return texture, image
def identify(save=True, force_input=False, from_file=False): """ Takes a picture with configured camera and identifies all of the faces in the picture Parameters: ----------- save (boolean): whether or not to add the captured image to the database from_file(boolean): whether or not expect a filename instead of taking a picture Returns: -------- names (list) the list of the name of each person in the picture """ if not from_file: img = ImageLoader.get_img_from_camera() dets = ImageLoader.find_faces(img) descs = ImageLoader.find_descriptors(img, dets) else: filepath = input('Please enter the location (filepath) of the image: ') img = ImageLoader.get_img_from_file(filepath) dets = ImageLoader.find_faces(img) descs = ImageLoader.find_descriptors(img, dets) names = ImageCompare.compare_faces(descs, database.database, threshold=0.4) if save: if len(descs) > 1: print("Cannot add multiple people at once.") elif len(descs) < 1: print("There's no one there!") else: if force_input: database.add_image(descs[0]) else: database.add_image(descs[0], name=names[0]) draw_faces(dets, names, img) return names
path = "./SavedModels/" print("Following are the list of Files: ") files = os.listdir(path) filecount = 0 for i in files: print(filecount+1, ")" , i , "\n") filecount+=1 print("Enter the number of the Model which you want to load") num = int(input()) #print("You chose " , files[num-1] ) filehandler = open(path+files[num-1], 'rb') nn = (pickle.load(filehandler)) training_data, validation_data, test_data = ImageLoader.load_data_wrapper() for (x, y) in test_data: print(np.argmax(nn.feedforward(x))) new_im = Image.fromarray((x*255).reshape(28,28)) new_im.resize((100,100)).show() #print(len(x)) input("Press Enter to go next") new_im.close()
F=f(H) fac=255/np.max(F) F=F*fac F.astype(int) print F cv2.imshow("skin",blur) cv2.waitKey(0) def HairRemoval(img): gray=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY) hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV) H,S,V=cv2.split(hsv) lower=120 upper=255 hairs = cv2.adaptiveThreshold(H,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,5,2) cv2.imshow("ip2",hairs) #cv2.imshow("res",res) cv2.waitKey(0) Kernal=cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(3,3)) print(Kernal) #return res IL=ImageLoader.Image_Loader() images=IL.NextSet() for image in images: #res=FeatureDetect(image) cv2.imshow("ip",image) Skin(image) #cv2.imshow("res",res) cv2.waitKey(0) cv2.destroyAllWindows()
non_trainable_count = np.sum( [K.count_params(p) for p in set(model.non_trainable_weights)]) total_memory = 4.0 * batch_size * (shapes_mem_count + trainable_count + non_trainable_count) gbytes = np.round(total_memory / (1024.0**3), 3) return gbytes K.set_image_dim_ordering('th') # fix random seed for reproducibility seed = 7 numpy.random.seed(seed) # load data #(X_train, y_train), (X_test, y_test) = mnist.load_data() (X_train, y_train) = ImageLoader.loadTrainingImages() (X_test, y_test) = ImageLoader.loadTestImages() # reshape to be [samples][pixels][width][height] wDim = X_train.shape[1] hDim = X_train.shape[2] X_train = X_train.reshape(X_train.shape[0], 3, wDim, hDim).astype('float32') X_test = X_test.reshape(X_test.shape[0], 3, wDim, hDim).astype('float32') # normalize inputs from 0-255 to 0-1 X_train = X_train / 255 X_test = X_test / 255 # one hot encode outputs y_train = np_utils.to_categorical(y_train) y_test = np_utils.to_categorical(y_test) num_classes = y_test.shape[1]
## Scale 3 # avg psnr on trainset : 25.90 db # avg psnr on testset : 25.75 db import CONST import numpy as np import math import ImageLoader import batch_manager import Image from scipy import ndimage from compute_psnr import compute_psnr dset_train, dset_test, dset_full_gt, dset_full_low = ImageLoader.ImageLoad() ## Batch Manager Instantiation BM = batch_manager.BatchManager() BM.init(dset_train, dset_test) mse_sum = 0 psnr = 0 ## Calculate PSNR, MSE of BICUBIC # bic_batch = BM.testsample() # nTBATCH = np.shape(bic_batch)[1] # for i in xrange(nTBATCH): # tmp_bic = bic_batch[1][i,:,:,0] - bic_batch[0][i,:,:,0] # mse = np.mean( np.square(tmp_bic) ) # mse_sum = mse_sum + mse # psnr = psnr + 20*math.log10(1.0/math.sqrt(mse+1e-10) ) #
def _readMask(self): img = ImageLoader.load( self._maskTextureName )#self.maskTexture.image#.convert("RGBA") imgstring = img.convert("RGBA").tostring("raw", "RGBA", 0, 1) buffer = array.array('B', imgstring ) self.maskWidth, self.maskHeight = img.size self.mask = buffer
from keras.utils import np_utils import ImageLoader as loader X_train, X_test, y_train, y_test = loader.loadImages() new_X_train = X_train.astype('float32') new_X_test = X_test.astype('float32') new_X_train /= 255 new_X_test /= 255 new_y_train = np_utils.to_categorical(y_train) new_y_test = np_utils.to_categorical(y_test) from keras.models import Sequential from keras.layers import Dense, Dropout, Flatten from keras.layers.convolutional import Conv2D, MaxPooling2D from keras.optimizers import SGD from keras.constraints import maxnorm model = Sequential() model.add(Conv2D(32, (3, 3), input_shape=(128, 128, 3), activation='relu', padding='same', kernel_constraint=maxnorm(3))) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Flatten()) model.add(Dense(512, activation='relu', kernel_constraint=maxnorm(3))) model.add(Dropout(0.5)) model.add(Dense(6, activation='softmax', kernel_constraint=maxnorm(3))) model.compile(loss='categorical_crossentropy', optimizer=SGD(lr=0.05), metrics=['accuracy']) model.fit(new_X_train, new_y_train, epochs=20, batch_size=32) import h5py model.save('Trained Model_SeeFood_new.h5')
fingerfile = '/remote/filer.gx/home.active/dyambay/Bob/FingerprintRec/CrossmatchTrain_Live2.txt' f = open(fingerfile, 'r') data_files = f.readlines() galnames = data_files[0:2] probenames = data_files[2:4] galDic,probeDic = ft.createDic(galnames, probenames) #for x in range(1): for x in range(0,1): #for x in range(len(data_files)): if x<=1: dict_type = 1 elif x>1: dict_type = 2 toLoad = data_files[x][0:len(data_files[x])-1] #loads selected image into an array loadedImage, height, width = IL.loadImage(toLoad) #Crops image around center point size = size+16 CroppedPrint, errors = cr.crop(height, width, loadedImage, errors, toLoad, size) cropped = np.asarray(CroppedPrint, dtype='uint8') #savetxt = '/idiap/home/dyambay/Bob/Saved Images/CropImage'+str(x)+'.jpg' #bob.io.save(cropped,savetxt) #takes binary image and calculates the orientation field and then extracts ROI based on it orientation = frp.orientation(CroppedPrint) np.savetxt('/remote/filer.gx/home.active/dyambay/Bob/FingerprintRec/orient' + str(x) +'.txt',orientation) '''#uses histogram equalization to enhance the fingerprint image enhancedImage = frp.hist_equal(CroppedPrint) #savetxt = '/idiap/home/dyambay/Bob/Saved Images/enhanceImage'+str(x)+'.jpg'
#C:\Users\Public\Skin Treatment DIP_ML\code from theano import * from theano import tensor as T import numpy as np import cv2 import ImageLoader rad=np.random row=0 col=0 IL=ImageLoader.Neural_Image_Loader() def get_Image(): global row,col img,mask=IL.NextSet() row,col,chann=img.shape x=img.reshape((row*col),chann) y=mask.reshape((row*col)) yp1=y>128 yp2=y<129 yp1=yp1*1 yp2=yp2*1 yp=np.array([yp1,yp2]) yp=yp.transpose() return x,yp Xi,Yi=get_Image() print(Xi,Yi) rng = np.random.RandomState(1234) #network sizes training_steps=1000 #Data input vaiable
def main(): with open("urls.xml", "rt") as urls: for url in urls: ImageLoader.load(url, "images") ImageAnalyzer.sortImages("images")
from ImageLoader import * Width = 416 #Width of network's input image Height = 416 #Height of network's input image # Give the configuration and weight files for the model and load the network using them. modelConfiguration = "yolov3.cfg" modelWeights = "yolov3.weights" net = cv.dnn.readNetFromDarknet(modelConfiguration, modelWeights) net.setPreferableBackend(cv.dnn.DNN_BACKEND_OPENCV) net.setPreferableTarget(cv.dnn.DNN_TARGET_CPU) device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') dataset = ImageLoader() loader = DataLoader(dataset, batch_size=256, shuffle=False, num_workers=(8 if device == "cuda" else 0)) for idx, (img_names, img_IDs) in enumerate(loader): start = time.time() img_IDs = img_IDs.numpy() imgs = [] for f in img_names: f = os.path.join('data/images/train2014/', f) if not os.path.isfile(f): print("Input image file ", args.image, " doesn't exist") sys.exit(1) img = cv.imread(f)