Bounding overwatch

About: Bounding overwatch is a research topic. Over the lifetime, 966 publications have been published within this topic receiving 15156 citations.

Is attention to bounding boxes all you need for pedestrian action prediction

Abstract: The human driver is no longer the only one concerned with the complexity of the driving scenarios. Autonomous vehicles (AV) are similarly becoming involved in the process. Nowadays, the development of AV in urban places underpins essential safety concerns for vulnerable road users (VRUs) such as pedestrians. Therefore, to make the roads safer, it is critical to classify and predict their future behavior. In this paper, we present a framework based on multiple variations of the Transformer models to reason attentively about the dynamic evolution of the pedestrians' past trajectory and predict its future actions of crossing or not crossing the street. We proved that using only bounding-boxes as input to our model can outperform the previous state-of-the-art models and reach a prediction accuracy of 91% and an F1-score of 0.83 on the PIE dataset up to two seconds ahead in the future. In addition, we introduced a large-size simulated dataset (CP2A) using CARLA for action prediction. Our model has similarly reached high accuracy (91 %) and F1-score (0.91) on this dataset. Interestingly, we showed that pre-training our Transformer model on the simulated dataset and then fine-tuning it on the real dataset can be very effective for the action prediction task.

A Restricted Dual Peaceman-Rachford Splitting Method for a Strengthened DNN Relaxation for QAP

Abstract: Splitting methods in optimization arise when one can divide an optimization problem into two or more simpler subproblems. They have proven particularly successful for relaxations of problems involving discrete variables. We revisit and strengthen splitting methods for solving doubly nonnegative relaxations of the particularly difficult, NP-hard quadratic assignment problem. We use a modified restricted contractive splitting method approach. In particular, we show how to exploit redundant constraints in the subproblems. Our strengthened bounds exploit these new subproblems and new dual multiplier estimates to improve on the bounds and convergence results in the literature. Summary of Contribution: In our paper, we consider the quadratic assignment problem (QAP). It is one of the fundamental combinatorial optimization problems in the fields of optimization and operations research and includes many fundamental applications. We revisit and strengthen splitting methods for solving doubly nonnegative (DNN) relaxation of the QAP. We use a modified restricted contractive splitting method. We obtain strengthened bounds from improved lower and upper bounding techniques, and in fact, we solve many of these NP-hard problems to (provable) optimality, thus illustrating both the strength of the DNN relaxation and our new bounding techniques.

Lagrangian Dual Decision Rules for Multistage Stochastic Mixed Integer Programming

Abstract: Multistage stochastic programs can be approximated by restricting policies to follow decision rules. Directly applying this idea to problems with integer decisions is difficult because of the need for decision rules that lead to integral decisions. In this work, we introduce Lagrangian dual decision rules (LDDRs) for multistage stochastic mixed integer programming (MSMIP) which overcome this difficulty by applying decision rules in a Lagrangian dual of the MSMIP. We propose two new bounding techniques based on stagewise (SW) and nonanticipative (NA) Lagrangian duals where the Lagrangian multiplier policies are restricted by LDDRs. We demonstrate how the solutions from these duals can be used to drive primal policies. Our proposal requires fewer assumptions than most existing MSMIP methods. We compare the theoretical strength of the restricted duals and show that the restricted NA dual can provide relaxation bounds at least as good as the ones obtained by the restricted SW dual. In our numerical study, we observe that the proposed LDDR approaches yield significant optimality gap reductions compared to existing general-purpose bounding methods for MSMIP problems.

Lung Cancer Detection using Deep Learning Approach CNN

Abstract: Thoracic radiography (chest X-ray) is a low-cost scientific imaging approach that is quite successful. However, because to a scarcity of skilled radiologists, the technique’s utility is severely limited. Even recent Deep Learning-based solutions sometimes require a lot of supervision to educate such systems, such as annotated bounding boxes, which is difficult to harvest on a large scale. This study recommends that frontal thoracic X-rays be classified and forecasted using a modified model called MobileNet V2. Every year, Computed Tomography (CT) should save a huge number of lives by finding most tumors early on. However, radiologists confront a significant task in analyzing many these images, and they frequently suffer from observer fatigue, which can affect their performance. As a result, it is necessary to read, identify, and consider CT images quickly. Using the NIH Chest-Xray-14 database, the overall performance of this technique is compared to the current modern-day pathology classification algorithms. Inconsistencies in classifiers were originally investigated using the Area Under the receiver operating characteristic Curve (AUC) data. Overall, the obtained result has a wide range, with an AUC of 0.811 and an accuracy of more than 90%. It is concluded that resampling the dataset improves the model’s performance significantly. The goal is to design a model that could be taught, as well as modified units that used less computational energy and could be used in smaller IoT devices.