Super resolution model test

The following code was used in testing super resolution model.

orig_img = Image.open("midge_header_2.jpg")
img = resizeimage.resize_cover(orig_img, [224,224], validate=False)
img_ycbcr = img.convert('YCbCr')
img_y_0, img_cb, img_cr = img_ycbcr.split()
img_ndarray = np.asarray(img_y_0)

img_4 = np.expand_dims(np.expand_dims(img_ndarray, axis=0), axis=0)
img_5 = img_4.astype(np.float32) / 255.0


ort_session = onnxruntime.InferenceSession("model/super-resolution-10/super_resolution/super_resolution.onnx")
ort_inputs = {ort_session.get_inputs()[0].name: img_5} 
ort_outs = ort_session.run(None, ort_inputs)
img_out_y = ort_outs[0]

img_out_y = Image.fromarray(np.uint8((img_out_y[0] * 255.0).clip(0, 255)[0]), mode='L')
# get the output image follow post-processing step from PyTorch implementation
final_img = Image.merge(
    "YCbCr", [
        img_out_y,
        img_cb.resize(img_out_y.size, Image.BICUBIC),
        img_cr.resize(img_out_y.size, Image.BICUBIC),
    ]).convert("RGB")
#plt.imshow(final_img)
plt.figure(figsize=(20, 10))

images = [orig_img, final_img]
titles = ['LR', f'SR (x{final_img.size[0] // orig_img.size[0]})']

for i, (img, title) in enumerate(zip(images, titles)):
   plt.subplot(1, 2, i+1)
   plt.imshow(img)
   plt.title(title)
   plt.xticks([])
   plt.yticks([])
plt.savefig('sample_detection.png', bbox_inches='tight')

Not very sure what parameter to modify.
But the output image was not full image as original and resolution is not good. The output image is blur.
What could be the problem?

You can see input and output image here

The test image to test is here.

Start by running the provided Python code with midge_header_2.jpg and the model at model/super-resolution-10/super_resolution/super_resolution.onnx. Compare the generated sample_detection.png with the linked input and output, focusing on the resize, YCbCr conversion, model input, and output post-processing. Done means identifying the parameter or processing step responsible for the cropped or blurred result.