![](https://csdnimg.cn/release/download_crawler_static/9231595/bg4.jpg)
4 Saurabh Gupta, Ross Girshick, Pablo Arbel´aez, Jitendra Malik
two approaches that combines features from [18] with the learning framework
from [9]. Specifically, we add the following features.
Normal Gradients: We compute normal gradients at two scales (corresponding
to fitting a local plane in a half-disk of radius 3 and 5 pixels), and use these as
additional gradient maps.
Geocentric Pose: We compute a per pixel height above ground and angle
with gravity (using the algorithms we proposed in [18]. These features allow the
decision trees to exploit additional regularities, for example that the brightness
edges on the floor are not as important as brightness edges elsewhere.
Richer Appearance: We observe that the NYUD2 dataset has limited ap-
pearance variation (since it only contains images of indoor scenes). To make the
model generalize better, we add the soft edge map produced by running the
RGB edge detector of [9] (which is trained on BSDS) on the RGB image.
2.2 Candidate Ranking
From the improved contour signal, we obtain object proposals by generalizing
MCG to RGB-D images. MCG for RGB images [1] uses simple features based
on the color image and the region shape to train a random forest regressors to
rank the object proposals. We follow the same paradigm, but propose additional
geometric features computed on the depth image within each proposal. We com-
pute: (1) the mean and standard deviation of the disparity, height above ground,
angle with gravity, and world (X, Y, Z) coordinates of the points in the region;
(2) the region’s (X, Y, Z) extent; (3) the region’s minimum and maximum height
above ground; (4) the fraction of pixels on vertical surfaces, surfaces facing up,
and surfaces facing down; (5) the minimum and maximum standard deviation
along a direction in the top view of the room. We obtain 29 geometric features
for each region in addition to the 14 from the 2D region shape and color image
already computed in [1]. Note that the computation of these features for a region
decomposes over superpixels and can be done efficiently by first computing the
first and second order moments on the superpixels and then combining them
appropriately.
2.3 Results
We now present results for contour detection and candidate ranking. We work
with the NYUD2 dataset and use the standard split of 795 training images and
654 testing images (we further divide the 795 images into a training set of 381
images and a validation set of 414 images). These splits are carefully selected
such that images from the same scene are only in one of these sets.
Contour detection: To measure performance on the contour detection task,
we plot the precision-recall curve on contours in Figure 3 and report the stan-
dard maximum F-measure metric (F
max
) in Table 1. We start by comparing the
performance of [18] (Gupta et al. CVPR [RGBD]) and Doll´ar et al. (SE [RGBD])
[9]. We see that both these contour detectors perform comparably in terms of