fix gaussian_blur

core/leras/ops/__init__.py

@@ -212,7 +212,9 @@ def gaussian_blur(input, radius=2.0):
         return np.exp(-(float(x) - float(mu)) ** 2 / (2 * sigma ** 2))
 
     def make_kernel(sigma):
-        kernel_size = max(3, int(2 * 2 * sigma + 1))
+        kernel_size = max(3, int(2 * 2 * sigma))
+        if kernel_size % 2 == 0:
+            kernel_size += 1
         mean = np.floor(0.5 * kernel_size)
         kernel_1d = np.array([gaussian(x, mean, sigma) for x in range(kernel_size)])
         np_kernel = np.outer(kernel_1d, kernel_1d).astype(np.float32)
@@ -403,3 +405,69 @@ def tf_suppress_lower_mean(t, eps=0.00001):
     q = q * (t/eps)
     return q
 """
+
+
+
+def _get_pixel_value(img, x, y):
+    shape = tf.shape(x)
+    batch_size = shape[0]
+    height = shape[1]
+    width = shape[2]
+
+    batch_idx = tf.range(0, batch_size)
+    batch_idx = tf.reshape(batch_idx, (batch_size, 1, 1))
+    b = tf.tile(batch_idx, (1, height, width))
+
+    indices = tf.stack([b, y, x], 3)
+    
+    return tf.gather_nd(img, indices)
+    
+def bilinear_sampler(img, x, y):
+    H = tf.shape(img)[1]
+    W = tf.shape(img)[2]
+    H_MAX = tf.cast(H - 1, tf.int32)
+    W_MAX = tf.cast(W - 1, tf.int32)
+
+    # grab 4 nearest corner points for each (x_i, y_i)
+    x0 = tf.cast(tf.floor(x), tf.int32)
+    x1 = x0 + 1
+    y0 = tf.cast(tf.floor(y), tf.int32)
+    y1 = y0 + 1
+
+    # clip to range [0, H-1/W-1] to not violate img boundaries
+    x0 = tf.clip_by_value(x0, 0, W_MAX)
+    x1 = tf.clip_by_value(x1, 0, W_MAX)
+    y0 = tf.clip_by_value(y0, 0, H_MAX)
+    y1 = tf.clip_by_value(y1, 0, H_MAX)
+
+    # get pixel value at corner coords
+    Ia = _get_pixel_value(img, x0, y0)
+    Ib = _get_pixel_value(img, x0, y1)
+    Ic = _get_pixel_value(img, x1, y0)
+    Id = _get_pixel_value(img, x1, y1)
+
+    # recast as float for delta calculation
+    x0 = tf.cast(x0, tf.float32)
+    x1 = tf.cast(x1, tf.float32)
+    y0 = tf.cast(y0, tf.float32)
+    y1 = tf.cast(y1, tf.float32)
+
+    # calculate deltas
+    wa = (x1-x) * (y1-y)
+    wb = (x1-x) * (y-y0)
+    wc = (x-x0) * (y1-y)
+    wd = (x-x0) * (y-y0)
+
+    # add dimension for addition
+    wa = tf.expand_dims(wa, axis=3)
+    wb = tf.expand_dims(wb, axis=3)
+    wc = tf.expand_dims(wc, axis=3)
+    wd = tf.expand_dims(wd, axis=3)
+
+    # compute output
+    out = tf.add_n([wa*Ia, wb*Ib, wc*Ic, wd*Id])
+
+    return out
+    
+nn.bilinear_sampler = bilinear_sampler
+