Support Python3 for face detection evalutation (#1319)

fluid/face_detection/pyramidbox.py

@@ -427,6 +427,7 @@ class PyramidBox(object):
             overlap_threshold=0.35,
             neg_overlap=0.35)
         loss = fluid.layers.reduce_sum(loss)
+        loss.persistable = True
         return loss
 
     def train(self):

fluid/face_detection/train.py

@@ -189,13 +189,13 @@ def train(args, config, train_params, train_file_list):
                 fetch_vars = [np.mean(np.array(v)) for v in fetch_vars]
                 if batch_id % 10 == 0:
                     if not args.use_pyramidbox:
-                        print("Pass {0}, batch {1}, loss {2}, time {3}".format(
+                        print("Pass {:d}, batch {:d}, loss {:.6f}, time {:.5f}".format(
                             pass_id, batch_id, fetch_vars[0],
                             start_time - prev_start_time))
                     else:
-                        print("Pass {0}, batch {1}, face loss {2}, " \
-                              "head loss {3}, " \
-                              "time {4}".format(pass_id,
+                        print("Pass {:d}, batch {:d}, face loss {:.6f}, " \
+                              "head loss {:.6f}, " \
+                              "time {:.5f}".format(pass_id,
                                batch_id, fetch_vars[0], fetch_vars[1],
                                start_time - prev_start_time))
             if pass_id % 1 == 0 or pass_id == epoc_num - 1:

fluid/face_detection/widerface_eval.py

@@ -82,9 +82,6 @@ def save_widerface_bboxes(image_path, bboxes_scores, output_dir):
     image_name = image_path.split('/')[-1]
     image_class = image_path.split('/')[-2]
 
-    image_name = image_name.encode('utf-8')
-    image_class = image_class.encode('utf-8')
-
     odir = os.path.join(output_dir, image_class)
     if not os.path.exists(odir):
         os.makedirs(odir)

fluid/object_detection/train.py

@@ -38,7 +38,8 @@ train_parameters = {
         "batch_size": 64,
         "lr": 0.001,
         "lr_epochs": [40, 60, 80, 100],
-        "lr_decay": [1, 0.5, 0.25, 0.1, 0.01]
+        "lr_decay": [1, 0.5, 0.25, 0.1, 0.01],
+        "ap_version": '11point',
     },
     "coco2014": {
         "train_images": 82783,
@@ -47,7 +48,8 @@ train_parameters = {
         "batch_size": 64,
         "lr": 0.001,
         "lr_epochs": [12, 19],
-        "lr_decay": [1, 0.5, 0.25]
+        "lr_decay": [1, 0.5, 0.25],
+        "ap_version": 'integral', # should use eval_coco_map.py to test model
     },
     "coco2017": {
         "train_images": 118287,
@@ -56,7 +58,8 @@ train_parameters = {
         "batch_size": 64,
         "lr": 0.001,
         "lr_epochs": [12, 19],
-        "lr_decay": [1, 0.5, 0.25]
+        "lr_decay": [1, 0.5, 0.25],
+        "ap_version": 'integral', # should use eval_coco_map.py to test model
     }
 }
 
@@ -77,6 +80,7 @@ def optimizer_setting(train_params):
 def build_program(main_prog, startup_prog, train_params, is_train):
     image_shape = train_params['image_shape']
     class_num = train_params['class_num']
+    ap_version = train_params['ap_version']
     with fluid.program_guard(main_prog, startup_prog):
         py_reader = fluid.layers.py_reader(
             capacity=64,
@@ -97,7 +101,6 @@ def build_program(main_prog, startup_prog, train_params, is_train):
 
                 nmsed_out = fluid.layers.detection_output(
                     locs, confs, box, box_var, nms_threshold=0.45)
-                with fluid.program_guard(main_prog):
                 loss = fluid.evaluator.DetectionMAP(
                     nmsed_out,
                     gt_label,
@@ -106,7 +109,7 @@ def build_program(main_prog, startup_prog, train_params, is_train):
                     class_num,
                     overlap_threshold=0.5,
                     evaluate_difficult=False,
-                        ap_version=args.ap_version)
+                    ap_version=ap_version)
     return py_reader, loss
 
 
@@ -230,7 +233,7 @@ def train(args,
                 loss_v = np.mean(np.array(loss_v))
                 every_epoc_loss.append(loss_v)
                 if batch_id % 20 == 0:
-                    print("Epoc {0}, batch {1}, loss {2}, time {3}".format(
+                    print("Epoc {:d}, batch {:d}, loss {:.6f}, time {:.5f}".format(
                         epoc_id, batch_id, loss_v, start_time - prev_start_time))
             end_time = time.time()
             total_time += end_time - start_time