# Update python scrypt

For multi scale training

darknet.py

@@ -111,6 +111,7 @@ def _process_batch(data, size_index):
     ious_reshaped = np.reshape(ious, [hw, num_anchors, len(cell_inds)])
     for i, cell_ind in enumerate(cell_inds):
         if cell_ind >= hw or cell_ind < 0:
+            print('cell inds size {}'.format(len(cell_inds)))
             print('cell over {} hw {}'.format(cell_ind, hw))
             continue
         a = anchor_inds[i]
@@ -168,6 +169,7 @@ def __init__(self):
         # linear
         out_channels = cfg.num_anchors * (cfg.num_classes + 5)
         self.conv5 = net_utils.Conv2d(c4, out_channels, 1, 1, relu=False)
+        self.global_average_pool = nn.AvgPool2d((1,1))
 
         # train
         self.bbox_loss = None
@@ -187,6 +189,7 @@ def forward(self, im_data, gt_boxes=None, gt_classes=None, dontcare=None, size_i
         cat_1_3 = torch.cat([conv1s_reorg, conv3], 1)
         conv4 = self.conv4(cat_1_3)
         conv5 = self.conv5(conv4)   # batch_size, out_channels, h, w
+        conv5 = self.global_average_pool(conv5)
 
         # for detection
         # bsize, c, h, w -> bsize, h, w, c -> bsize, h x w, num_anchors, 5+num_classes

datasets/imdb.py

@@ -2,6 +2,9 @@
 import PIL
 import numpy as np
 from multiprocessing import Pool
+from functools import partial
+import cfgs.config as cfg
+import cv2
 
 
 def mkdir(path, max_depth=3):
@@ -13,6 +16,12 @@ def mkdir(path, max_depth=3):
         os.mkdir(path)
 
 
+def image_resize(im, size_index):
+    w, h = cfg.multi_scale_inp_size[size_index]
+    im = cv2.resize(im, (w, h))
+    return im
+
+
 class ImageDataset(object):
     def __init__(self, name, datadir, batch_size, im_processor, processes=3, shuffle=True, dst_size=None):
         self._name = name
@@ -38,12 +47,13 @@ def __init__(self, name, datadir, batch_size, im_processor, processes=3, shuffle
         self.gen = None
         self._im_processor = im_processor
 
-    def next_batch(self):
+    def next_batch(self, size_index):
         batch = {'images': [], 'gt_boxes': [], 'gt_classes': [], 'dontcare': [], 'origin_im': []}
         i = 0
         while i < self.batch_size:
             try:
                 images, gt_boxes, classes, dontcare, origin_im = next(self.gen)
+                images = image_resize(images, size_index)
                 batch['images'].append(images)
                 batch['gt_boxes'].append(gt_boxes)
                 batch['gt_classes'].append(classes)
@@ -54,7 +64,7 @@ def next_batch(self):
                 indexes = np.arange(len(self.image_names), dtype=np.int)
                 if self._shuffle:
                     np.random.shuffle(indexes)
-                self.gen = self.pool.imap(self._im_processor,
+                self.gen = self.pool.imap(partial(self._im_processor, size_index=size_index),
                                           ([self.image_names[i], self.get_annotation(i), self.dst_size] for i in indexes),
                                           chunksize=self.batch_size)
                 self._epoch += 1