如何将我的 csv 文件像素数据转换为图像

Question

I am totally beginner in this field. I started working with neural network for image classification purpose. My question is I loaded one row through panda. now I want to see that image like from which category it is. it has label 0. so how i can convert that pixels values into image.

import matplotlib.pyplot as plt
import tensorflow as tf
import numpy as np
import pandas as pd
import seaborn as sns
import cv2
import os

#import deep learning libraries
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.preprocessing import image

df = pd.read_csv(r"/content/drive/MyDrive/Colab Notebooks/fe/icml_face_data.csv", nrows = 1)
df.pixel

0    70 80 82 72 58 58 60 63 54 58 60 48 89 115 121...
Name: pixel, dtype: object

The above pixel values I am getting now in next step I want to convert this into image.

Answer 1

You can visualise an image using matplotlib ( plt.imshow ) or seaborn ( sns.heatmap ). Note that in all cases you'll probably want to change the colour map to something other than the default.

However, it looks like your image is stored in a vector, not a 2D matrix. You can reshape this using numpy ( np.reshape ) but you will need to know the original dimensions of your image for this transformation. You can use df.pixel.to_numpy() to make a numpy vector.