add resampling

inference_test.py

@@ -107,8 +107,14 @@ class TextProcessor:
         return final_text
 
 
-import cv2
-import numpy as np
+def bilinear(img):
+    h, w = img.shape[:2]
+    # Create identity maps
+    map_x, map_y = np.meshgrid(np.arange(w, dtype=np.float32),
+                               np.arange(h, dtype=np.float32))
+    # Remap with INTER_LINEAR
+    return cv2.remap(img, map_x, map_y, interpolation=cv2.INTER_LINEAR)
+
 
 def adjust_gamma(image, gamma=1.0, alpha=1.0, beta=0):
     """
@@ -137,15 +143,15 @@ def adjust_gamma(image, gamma=1.0, alpha=1.0, beta=0):
 def preprocess_image(image_path):
     # Load image in grayscale (1-channel)
     gray = cv2.imread(image_path, cv2.IMREAD_GRAYSCALE)
-
+    resampled = bilinear(gray) 
     # Convert grayscale to 3-channel RGB by duplicating the gray channel
-    img = adjust_gamma(cv2.cvtColor(gray, cv2.COLOR_GRAY2RGB),gamma=1.3,alpha=1.65,beta=-2) # perfect tunning
+    img = adjust_gamma(cv2.cvtColor(resampled, cv2.COLOR_GRAY2RGB),gamma=1.3,alpha=1.65,beta=-2) # perfect tunning
     mat_in = ncnn.Mat.from_pixels_resize(
         img, ncnn.Mat.PixelType.PIXEL_RGB, img.shape[1], img.shape[0], 224, 224
     )
     
-    mean_vals = [0.5, 0.5, 0.5] * 2
-    norm_vals = [1/255, 1/255, 1/255] * 2
+    mean_vals = [0.5, 0.5, 0.5] * 3
+    norm_vals = [1/255, 1/255, 1/255] * 3
     mat_in.substract_mean_normalize(mean_vals, norm_vals)
     return mat_in