Parse the MNIST images header

mnist.py

# --- Images Header
# 4x 32-bit big-endian integer
images_header = np.fromfile(images_file, dtype='>i4', count=4)

images_magic_number = images_header[0]
images_count = images_header[1]
row_count = images_header[2]
col_count = images_header[3]
pixel_count = row_count*col_count

print(f'magic number\t {images_magic_number}')
print(f'#images\t\t {images_count}')
print(f'#rows\t\t {row_count}')
print(f'#cols\t\t {col_count}')
print(f'#pixels\t\t {pixel_count} ({row_count}*{col_count})')

# --- Images Data
images_buffer = images_file.read(pixel_count * images_count)
images = np.frombuffer(images_buffer, dtype=np.uint8)
images = images.reshape(images_count, pixel_count)

 # --- Plot first image
img = images[0].reshape((28, 28))
plt.imshow(img, cmap='gray')
plt.show()