Remove FloatBox and IntBox

docs/conf.py

@@ -364,6 +364,7 @@ _DEPRECATED_NAMES = set([
     'dump_dataflow_to_lmdb',
     'dump_dataflow_to_tfrecord',
     'pyplot2img',
+    'IntBox', 'FloatBox',
 
     # renamed stuff:
     'DumpTensor',

examples/FasterRCNN/coco.py

@@ -8,7 +8,6 @@ from tabulate import tabulate
 import tqdm
 
 from tensorpack.utils import logger
-from tensorpack.utils.rect import FloatBox
 from tensorpack.utils.timer import timed_operation
 from tensorpack.utils.argtools import log_once
 
@@ -130,13 +129,15 @@ class COCODetection(object):
             x1, y1, w, h = obj['bbox']
             # bbox is originally in float
             # x1/y1 means upper-left corner and w/h means true w/h. This can be verified by segmentation pixels.
-            # But we do assume that (0.0, 0.0) is upper-left corner of the first pixel
-            box = FloatBox(float(x1), float(y1),
-                           float(x1 + w), float(y1 + h))
-            box.clip_by_shape([height, width])
+            # But we do make an assumption here that (0.0, 0.0) is upper-left corner of the first pixel
+
+            x1 = np.clip(float(x1), 0, width)
+            y1 = np.clip(float(y1), 0, height)
+            w = np.clip(float(x1 + w), 0, width) - x1
+            h = np.clip(float(y1 + h), 0, height) - y1
             # Require non-zero seg area and more than 1x1 box size
-            if obj['area'] > 1 and box.is_box() and box.area() >= 4:
-                obj['bbox'] = [box.x1, box.y1, box.x2, box.y2]
+            if obj['area'] > 1 and w > 0 and h > 0 and w * h >= 4:
+                obj['bbox'] = [x1, y1, x1 + w, y1 + h]
                 valid_objs.append(obj)
 
                 if add_mask:

examples/OpticalFlow/README.md

@@ -23,7 +23,7 @@ which is better than other TensorFlow implementations.
 
 ### Usage
 
-1. Download the pre-trained model:
+1. Download the pre-trained model (converted from caffe):
 
 ```bash
 wget http://models.tensorpack.com/OpticalFlow/flownet2.npz
@@ -36,7 +36,7 @@ wget http://models.tensorpack.com/OpticalFlow/flownet2-c.npz
 2. Run inference
 
 ```bash
-python flownet2.py
+python flownet2.py \
 	--left left.png --right right.png \
 	--load flownet2.npz --model flownet2
 ```

tensorpack/dataflow/imgaug/base.py

@@ -113,20 +113,20 @@ class ImageAugmentor(Augmentor):
     floating point images in range [0, 1] or [0, 255].
     """
     def augment_coords(self, coords, param):
-        return self._augment_coords(coords, param)
-
-    def _augment_coords(self, coords, param):
         """
         Augment the coordinates given the param.
-        By default, keeps coordinates unchanged.
+        By default, an augmentor keeps coordinates unchanged.
         If a subclass changes coordinates but couldn't implement this method,
         it should ``raise NotImplementedError()``.
 
         Args:
-            coords: Nx2 floating point nparray where each row is (x, y)
+            coords: Nx2 floating point numpy array where each row is (x, y)
         Returns:
             new coords
         """
+        return self._augment_coords(coords, param)
+
+    def _augment_coords(self, coords, param):
         return coords
 
 

tensorpack/utils/viz.py

@@ -8,7 +8,6 @@ import sys
 import io
 from .fs import mkdir_p
 from .argtools import shape2d
-from .rect import BoxBase, IntBox
 from .palette import PALETTE_RGB
 
 try:
@@ -367,8 +366,7 @@ def draw_boxes(im, boxes, labels=None, color=None):
     """
     Args:
         im (np.ndarray): a BGR image in range [0,255]. It will not be modified.
-        boxes (np.ndarray or list[BoxBase]): If an ndarray,
-            must be of shape Nx4 where the second dimension is [x1, y1, x2, y2].
+        boxes (np.ndarray): a numpy array of shape Nx4 where each row is [x1, y1, x2, y2].
         labels: (list[str] or None)
         color: a 3-tuple (in range [0, 255]). By default will choose automatically.
 
@@ -377,14 +375,7 @@ def draw_boxes(im, boxes, labels=None, color=None):
     """
     FONT = cv2.FONT_HERSHEY_SIMPLEX
     FONT_SCALE = 0.4
-    if isinstance(boxes, list):
-        arr = np.zeros((len(boxes), 4), dtype='int32')
-        for idx, b in enumerate(boxes):
-            assert isinstance(b, BoxBase), b
-            arr[idx, :] = [int(b.x1), int(b.y1), int(b.x2), int(b.y2)]
-        boxes = arr
-    else:
-        boxes = boxes.astype('int32')
+    boxes = np.asarray(boxes, dtype='int32')
     if labels is not None:
         assert len(labels) == len(boxes), "{} != {}".format(len(labels), len(boxes))
     areas = (boxes[:, 2] - boxes[:, 0] + 1) * (boxes[:, 3] - boxes[:, 1] + 1)
@@ -415,17 +406,18 @@ def draw_boxes(im, boxes, labels=None, color=None):
             if top_left[1] < 0:     # out of image
                 top_left[1] = box[3] - 1.3 * lineh
                 bottom_left[1] = box[3] - 0.3 * lineh
-            textbox = IntBox(int(top_left[0]), int(top_left[1]),
-                             int(top_left[0] + linew), int(top_left[1] + lineh))
-            textbox.clip_by_shape(im.shape[:2])
+            x1, y1 = int(top_left[0]), int(top_left[1])
+            x2, y2 = int(top_left[0] + linew), int(top_left[1] + lineh)
+            x1, x2 = [np.clip(x, 0, img.shape[1] - 1) for x in [x1, x2]]
+            y1, y2 = [np.clip(y, 0, img.shape[0] - 1) for y in [y1, y2]]
             if color is None:
                 # find the best color
-                mean_color = textbox.roi(im).mean(axis=(0, 1))
+                mean_color = im[y1:y2 + 1, x1:x2 + 1].mean(axis=(0, 1))
                 best_color_ind = (np.square(COLOR_CANDIDATES - mean_color) *
                                   COLOR_DIFF_WEIGHT).sum(axis=1).argmax()
                 best_color = COLOR_CANDIDATES[best_color_ind].tolist()
 
-            cv2.putText(im, label, (textbox.x1, textbox.y2),
+            cv2.putText(im, label, (x1, y2),
                         FONT, FONT_SCALE, color=best_color, lineType=cv2.LINE_AA)
         cv2.rectangle(im, (box[0], box[1]), (box[2], box[3]),
                       color=best_color, thickness=1)