Tracklet Proposal Network for Multi-Object Tracking on Point Clouds

This paper proposes the first tracklet proposal network, named PC-TCNN, for Multi-Object Tracking (MOT) on point clouds. Our pipeline first generates tracklet proposals, then refines these tracklets and associates them to generate long trajectories. Specifically, object proposal generation and motion regression are first performed on a point cloud sequence to generate tracklet candidates. Then, spatial-temporal features of each tracklet are exploited and their consistency is used to refine the tracklet proposal. Finally, the refined tracklets across multiple frames are associated to perform MOT on the point cloud sequence. The PC-TCNN significantly improves the MOT performance by introducing the tracklet proposal design. On the KITTI tracking benchmark, it attains an MOTA of 91.75%, outperforming all submitted results on the online leaderboard.

EmbedMask: Embedding Coupling for Instance Segmentation

Current instance segmentation methods can be categorized into segmentation-based methods and proposal-based methods. The former performs segmentation first and then does clustering, while the latter detects objects first and then predicts the mask for each object proposal. In this work, we propose a single-stage method, named EmbedMask, that unifies both methods by taking their advantages, so it can achieve good performance in instance segmentation and produce high-resolution masks in a high speed. EmbedMask introduces two newly defined embeddings for mask prediction, which are pixel embedding and proposal embedding. During training, we enforce the pixel embedding to be close to its coupled proposal embedding if they belong to the same instance. During inference, pixels are assigned to the mask of the proposal if their embeddings are similar. This mechanism brings several benefits. First, the pixel-level clustering enables EmbedMask to generate high-resolution masks and avoids the complicated two-stage mask prediction. Second, the existence of proposal embedding simplifies and strengthens the clustering procedure, so our method can achieve high speed and better performance than segmentation-based methods. Without any bell or whistle, EmbedMask outperforms the state-of-the-art instance segmentation method Mask R-CNN on the challenging COCO dataset, obtaining more detailed masks at a higher speed.

An Object Detection Framework Based on Deep Features and High-Quality Object Locations

Objection detection has long been a fundamental issue in computer vision. Despite being widely studied, it remains a challenging task in the current body of knowledge. Many researchers are eager to develop a more robust and efficient mechanism for object detection. In the extant literature, promising results are achieved by many novel approaches of object detection and classification. However, there is ample room to further enhance the detection efficiency. Therefore, this paper proposes an image object detection and classification, using a deep neural network (DNN) for based on high-quality object locations. The proposed method firstly derives high-quality class-independent object proposals (locations) through computing multiple hierarchical segments with super pixels. Next, the proposals were ranked by region score, i.e., several contours wholly enclosed in the proposed region. After that, the top-ranking object proposal was adopted for post-classification by the DNN. During the post-classification, the network extracts the eigenvectors from the proposals, and then maps the features with the softmax classifier, thereby determining the class of each object. The proposed method was found superior to traditional approaches through an evaluation on Pascal VOC 2007 Dataset.

A Region-Based Efficient Network for Accurate Object Detection

Object detection in images is an important task in image processing and computer vision. Many approaches are available for object detection. For example, there are numerous algorithms for object positioning and classification in images. However, the current methods perform poorly and lack experimental verification. Thus, it is a fascinating and challenging issue to position and classify image objects. Drawing on the recent advances in image object detection, this paper develops a region-baed efficient network for accurate object detection in images. To improve the overall detection performance, image object detection was treated as a twofold problem, involving object proposal generation and object classification. First, a framework was designed to generate high-quality, class-independent, accurate proposals. Then, these proposals, together with their input images, were imported to our network to learn convolutional features. To boost detection efficiency, the number of proposals was reduced by a network refinement module, leaving only a few eligible candidate proposals. After that, the refined candidate proposals were loaded into the detection module to classify the objects. The proposed model was tested on the test set of the famous PASCAL Visual Object Classes Challenge 2007 (VOC2007). The results clearly demonstrate that our model achieved robust overall detection efficiency over existing approaches using fewer or more proposals, in terms of recall, mean average best overlap (MABO), and mean average precision (mAP).

A Weakly Supervised Semantic Segmentation Network by Aggregating Seed Cues: The Multi-Object Proposal Generation Perspective

Weakly supervised semantic segmentation under image-level annotations is effectiveness for real-world applications. The small and sparse discriminative regions obtained from an image classification network that are typically used as the important initial location of semantic segmentation also form the bottleneck. Although deep convolutional neural networks (DCNNs) have exhibited promising performances for single-label image classification tasks, images of the real-world usually contain multiple categories, which is still an open problem. So, the problem of obtaining high-confidence discriminative regions from multi-label classification networks remains unsolved. To solve this problem, this article proposes an innovative three-step framework within the perspective of multi-object proposal generation. First, an image is divided into candidate boxes using the object proposal method. The candidate boxes are sent to a single-classification network to obtain the discriminative regions. Second, the discriminative regions are aggregated to obtain a high-confidence seed map. Third, the seed cues grow on the feature maps of high-level semantics produced by a backbone segmentation network. Experiments are carried out on the PASCAL VOC 2012 dataset to verify the effectiveness of our approach, which is shown to outperform other baseline image segmentation methods.

A Pedestrian Detection Algorithm Based on Score Fusion for Multi-LiDAR Systems

Pedestrian detection plays an essential role in the navigation system of autonomous vehicles. Multisensor fusion-based approaches are usually used to improve detection performance. In this study, we aimed to develop a score fusion-based pedestrian detection algorithm by integrating the data of two light detection and ranging systems (LiDARs). We first evaluated a two-stage object-detection pipeline for each LiDAR, including object proposal and fine classification. The scores from these two different classifiers were then fused to generate the result using the Bayesian rule. To improve proposal performance, we applied two features: the central points density feature, which acts as a filter to speed up the process and reduce false alarms; and the location feature, including the density distribution and height difference distribution of the point cloud, which describes an object’s profile and location in a sliding window. Extensive experiments tested in KITTI and the self-built dataset show that our method could produce highly accurate pedestrian detection results in real-time. The proposed method not only considers the accuracy and efficiency but also the flexibility for different modalities.

Visual Tracking of Small Unmanned Aerial Vehicles Based on Object Proposal Voting

In this paper, we propose a novel visual tracking method for unmanned aerial vehicles (UAVs) in aerial scenery. To track the UAVs robustly, we present a new object proposal method that can accurately determine the object regions that are likely to exist. The proposed object proposal method is robust to small objects and severe background clutter. For this, we vote on candidate areas of the object and increase or decrease the weight of the area accordingly. Thus, the method can accurately propose the object areas that can be used to track small-sized UAVs with the assumption that their motion is smooth over time. Experimental results verify that UAVs are accurately tracked even when they are very small and the background is complex. The proposed method qualitatively and quantitatively delivers state-of-the-art performance in comparison with conventional object proposal-based methods.