def __getitem__(self, index): # Select sample if self.input_size == 3: return ((np.asarray(self.data_im[index]) - self.mean_image) / self.std_image),self.exprs[index] elif self.input_size == 68: return np.asarray(heatmap_generator.getHeatMap(self.data_lm[index])), self.exprs[index] else: return np.vstack([((np.asarray(self.data_im[index]) - self.mean_image) / self.std_image),np.asarray(heatmap_generator.getHeatMap(self.data_lm[index]))]),self.exprs[index]
def __getitem__(self, index): # Select sample #image = 0.21*self.data_im[index][0,:,:]+0.72*self.data_im[index][1,:,:]+0.07*self.data_im[index][2,:,:] #mean = 0.21*self.mean_image[0,:,:]+0.72*self.mean_image[1,:,:]+0.07*self.mean_image[2,:,:] #std = 0.21*self.std_image[0,:,:]+0.72*self.std_image[1,:,:]+0.07*self.std_image[2,:,:] #print(np.asarray([(image-mean)/std]).shape) #print("*******",self.data_lm[index].shape) return np.asarray( (self.data_im[index] - self.mean_image) / self.std_image), np.asarray( heatmap_generator.getHeatMap( self.data_lm[index])), (self.exprs[index] if self.exprs[index] < 5 else 4)
def __getitem__(self, index): # Select sample return ((np.asarray(self.data_im[index]) - self.mean_image) / self.std_image), np.asarray(heatmap_generator.getHeatMap(self.data_lm[index])), self.exprs[index]