Pose Estimation of Object Categories in Videos Using Linear Programming

Abstract

In this paper we propose a method to consistently recover the pose of an object from a known class in a video sequence. As individual poses estimated from monocular images are rather noisy, we optimally aggregate pose evidence over all video frames. We construct a graph where nodes are values sampled from the pose posterior distributions computed by a continuous pose estimator in each frame of the sequence. We then find the globally optimum pose path through the graph that best explains the pose evidence for the whole sequence. As a result, we recover the correct object orientation at each frame even if single-frame pose evidence is sometimes inaccurate. We evaluate our approach on two publicly available car datasets, which encompass busy street scenarios and car races with significant changes in car orientation, blur and occlusions. We show that our method outperforms state-of-the-art approaches reducing the error by 40% on the challenging KITTI dataset.

SAMP: Shape and Motion Priors for 4D Vehicle Reconstruction

“…In recent years, also object-based alignment over multiple frames has been investigated for scene understanding in urban street scenes. Feni et al [7] estimate the yaw angle of vehicles from annotated 2D detections. Menze et al [14] model moving rigid objects for improving stereo scene flow estimation between two consecutive frames.…”

“…shape prior (7) In the following, we will explain the individual terms of the energy function in detail. Figure 4 shows an illustration of all the energy terms.…”

Continuous Pose Estimation with a Spatial Ensemble of Fisher Regressors

“…First, we evaluate our method by comparing with [30], as none of the two methods exploit temporal information, i.e., pose estimation is performed separately in each frame. Then, in order to compare with [31], we extend our model with a Linear Programming (LP) formulation to take temporal information into account, similarly to [7]. For this purpose, we exploit the posterior distribution delivered by our method trained on the first 10 sequences of the EPFL dataset.…”