In this paper, we present a novel approach, called Deep MANTA (Deep Many-Tasks), for many-task vehicle analysis from a given image. A robust convolutional network is introduced for simultaneous vehicle detection, part localization, visibility characterization and 3D dimension estimation. Its architecture is based on a new coarse-to-fine object proposal that boosts the vehicle detection. Moreover, the Deep MANTA network is able to localize vehicle parts even if these parts are not visible. In the inference, the network's outputs are used by a real time robust pose estimation algorithm for fine orientation estimation and 3D vehicle localization. We show in experiments that our method outperforms monocular state-of-the-art approaches on vehicle detection, orientation and 3D location tasks on the very challenging KITTI benchmark.

Figure 1: Qualitative results of PandaNet on JTA dataset [8] which consists in images with many people (up to 60), a large proportion of people at low resolution and many occlusion situations. Most of the previous 3D human pose estimation studies mainly focused on the single-person case or estimate 3D pose of few people at high resolution. In this paper, we propose an anchor-based and single-shot multi-person 3D pose estimation framework that allows the pose estimation of a large number of people at low resolution. Ground-truth translations and scales are used for visualisation.

Several modern deep single-stage object detectors are really effective for real time processing [19,20,21,22] but still remain less efficient than more complex ones [17,28]. The trade-off between model performances and computing speed is an important challenge, directly related to the learning process. In this paper, we propose a new way to efficiently learn a single shot detector providing a very good trade-off between these two factors (Figure 1). For this purpose, we introduce LapNet, an anchor based detector, trained end-to-end without any sampling strategy. Our approach focuses on two limitations of anchor based detector training: (1) the ambiguity of anchor to ground truth assignment and (2) the imbalance between classes and the imbalance between object sizes. More specifically, a new method to assign positive and negative anchors is proposed, based on a new overlapping function called "Per-Object Normalized Overlap" (PONO). This more flexible assignment can be self-corrected by the network itself to avoid the ambiguity between close objects. In the learning process, we also propose to automatically learn weights to balance classes and object sizes to efficiently manage sample imbalance. It allows to build a robust object detector avoiding multi-scale prediction, in a semantic segmentation spirit.

No abstract

International audienc