optimize image loading

common/backbones/imagenet_training/train_imagenet.py

@@ -199,7 +199,7 @@ def verify_with_image(model, input_shape):
     while True:
         img_file = input('Input image filename:')
         try:
-            img = Image.open(img_file)
+            img = Image.open(img_file).convert('RGB')
             resized_img = img.resize(input_shape, Image.BICUBIC)
         except:
             print('Open Error! Try again!')
@@ -215,10 +215,10 @@ def verify_with_image(model, input_shape):
             # show predict result on origin image
             img_array = np.asarray(img)
             cv2.putText(img_array, '{name}:{conf:.3f}'.format(name=result[0][0][1], conf=float(result[0][0][2])),
-                        (10,30),
+                        (10, 30),
                         cv2.FONT_HERSHEY_SIMPLEX,
                         fontScale=1,
-                        color=(255,0,0),
+                        color=(255, 0, 0),
                         thickness=2,
                         lineType=cv2.LINE_AA)
             Image.fromarray(img_array).show()

tools/dataset_converter/dataset_visualize.py

@@ -2,6 +2,7 @@
 # -*- coding: utf-8 -*-
 import os, sys, argparse
 import numpy as np
+from PIL import Image
 import cv2
 from tqdm import tqdm
 
@@ -19,7 +20,8 @@ def dataset_visualize(annotation_file, classes_path):
     for i, annotation_line in enumerate(annotation_lines):
         pbar.update(1)
         line = annotation_line.split()
-        image = cv2.imread(line[0])
+        image = Image.open(line[0]).convert('RGB')
+        image = np.array(image, dtype='uint8')
         boxes = np.array([np.array(list(map(int, box.split(',')))) for box in line[1:]])
 
         classes = boxes[:, -1]
@@ -31,13 +33,16 @@ def dataset_visualize(annotation_file, classes_path):
         # show image file info
         image_file_name = os.path.basename(line[0])
         cv2.putText(image, image_file_name+'({}/{})'.format(i+1, len(annotation_lines)),
-                    (3,15),
+                    (3, 15),
                     cv2.FONT_HERSHEY_PLAIN,
                     fontScale=1,
-                    color=(0,0,255),
+                    color=(255, 0, 0),
                     thickness=1,
                     lineType=cv2.LINE_AA)
 
+        # convert to BGR for cv2.imshow
+        image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)
+
         cv2.namedWindow("Image", 0)
         cv2.imshow("Image", image)
         keycode = cv2.waitKey(0) & 0xFF

tools/evaluation/validate_yolo.py

@@ -27,7 +27,7 @@ def validate_yolo_model(model_path, image_file, anchors, class_names, model_inpu
     custom_object_dict = get_custom_objects()
     model = load_model(model_path, compile=False, custom_objects=custom_object_dict)
 
-    img = Image.open(image_file)
+    img = Image.open(image_file).convert('RGB')
     image = np.array(img, dtype='uint8')
     image_data = preprocess_image(img, model_input_shape)
     #origin image shape, in (height, width) format
@@ -63,7 +63,7 @@ def validate_yolo_model_tflite(model_path, image_file, anchors, class_names, eli
     if input_details[0]['dtype'] == np.float32:
         floating_model = True
 
-    img = Image.open(image_file)
+    img = Image.open(image_file).convert('RGB')
     image = np.array(img, dtype='uint8')
 
     height = input_details[0]['shape'][1]
@@ -114,7 +114,7 @@ def validate_yolo_model_mnn(model_path, image_file, anchors, class_names, elim_g
     model_input_shape = (height, width)
 
     # prepare input image
-    img = Image.open(image_file)
+    img = Image.open(image_file).convert('RGB')
     image = np.array(img, dtype='uint8')
     image_data = preprocess_image(img, model_input_shape)
     #origin image shape, in (height, width) format
@@ -280,7 +280,7 @@ def load_pb_graph(model_path):
     batch, height, width, channel = image_input.shape
     model_input_shape = (int(height), int(width))
 
-    img = Image.open(image_file)
+    img = Image.open(image_file).convert('RGB')
     image = np.array(img, dtype='uint8')
     image_data = preprocess_image(img, model_input_shape)
     #origin image shape, in (height, width) format
@@ -326,7 +326,7 @@ def validate_yolo_model_onnx(model_path, image_file, anchors, class_names, elim_
     model_input_shape = (height, width)
 
     # prepare input image
-    img = Image.open(image_file)
+    img = Image.open(image_file).convert('RGB')
     image = np.array(img, dtype='uint8')
     image_data = preprocess_image(img, model_input_shape)
     #origin image shape, in (height, width) format

tools/misc/rotate_gridmask_test.py

@@ -14,13 +14,13 @@
 
 def main():
     # load some VOC2007 images for test
-    #image = Image.open("000001.jpg")
+    #image = Image.open("000001.jpg").convert('RGB')
     #boxes = np.array([[48,240,195,371,11],[8,12,352,498,14]])
 
-    image = Image.open("000004.jpg")
+    image = Image.open("000004.jpg").convert('RGB')
     boxes = np.array([[13,311,84,362,6],[362,330,500,389,6],[235,328,334,375,6],[175,327,252,364,6],[139,320,189,359,6],[108,325,150,353,6],[84,323,121,350,6]])
 
-    #image = Image.open("000010.jpg")
+    #image = Image.open("000010.jpg").convert('RGB')
     #boxes = np.array([[87,97,258,427,12],[133,72,245,284,14]])
 
 

yolo.py

@@ -313,7 +313,7 @@ def detect_img(yolo):
     while True:
         img = input('Input image filename:')
         try:
-            image = Image.open(img)
+            image = Image.open(img).convert('RGB')
         except:
             print('Open Error! Try again!')
             continue

yolo3/data.py

@@ -12,7 +12,7 @@
 def get_ground_truth_data(annotation_line, input_shape, augment=True, max_boxes=100):
     '''random preprocessing for real-time data augmentation'''
     line = annotation_line.split()
-    image = Image.open(line[0])
+    image = Image.open(line[0]).convert('RGB')
     image_size = image.size
     model_input_size = input_shape[::-1]
     boxes = np.array([np.array(list(map(int, box.split(',')))) for box in line[1:]])