Multi-modal sensor data fusion takes advantage of complementary or reinforcing information from each sensor and can boost overall performance in applications such as scene classification and target detection. This paper presents a new method for fusing multi-modal and multi-resolution remote sensor data without requiring pixel-level training labels, which can be difficult to obtain. Previously, we developed a Multiple Instance Multi-Resolution Fusion (MIMRF) framework that addresses label uncertainty for fusion, but it can be slow to train due to the large search space for the fuzzy measures used to integrate sensor data sources. We propose a new method based on binary fuzzy measures, which reduces the search space and significantly improves the efficiency of the MIMRF framework. We present experimental results on synthetic data and a real-world remote sensing detection task and show that the proposed MIMRF-BFM algorithm can effectively and efficiently perform multi-resolution fusion given remote sensing data with uncertainty.

Accurate prediction of pedestrian crossing behaviors by autonomous vehicles can significantly improve traffic safety. Existing approaches often model pedestrian behaviors using trajectories or poses but do not offer a deeper semantic interpretation of a person's actions or how actions influence a pedestrian's intention to cross in the future. In this work, we follow the neuroscience and psychological literature to define pedestrian crossing behavior as a combination of an unobserved inner will (a probabilistic representation of binary intent of crossing vs. not crossing) and a set of multi-class actions (e.g., walking, standing, etc.). Intent generates actions, and the future actions in turn reflect the intent. We present a novel multi-task network that predicts future pedestrian actions and uses predicted future action as a prior to detect the present intent and action of the pedestrian. We also designed an attention relation network to incorporate external environmental contexts thus further improve intent and action detection performance. We evaluated our approach on two naturalistic driving datasets, PIE and JAAD, and extensive experiments show significantly improved and more explainable results for both intent detection and action prediction over state-of-the-art approaches. Our code is available at: https://github.com/umautobots/pedestrian_intent_action_detection

In remote sensing, each sensor can provide complementary or reinforcing information. It is valuable to fuse outputs from multiple sensors to boost overall performance. Previous supervised fusion methods often require accurate labels for each pixel in the training data. However, in many remote-sensing applications, pixel-level labels are difficult or infeasible to obtain. In addition, outputs from multiple sensors often have different resolutions or modalities. For example, rasterized hyperspectral imagery (HSI) presents data in a pixel grid while airborne light detection and ranging (LiDAR) generates dense 3-D point clouds. It is often difficult to directly fuse such multimodal, multiresolution data. To address these challenges, we present a novel multiple instance multiresolution fusion (MIMRF) framework that can fuse multiresolution and multimodal sensor outputs while learning from automatically generated, imprecisely labeled data. Experiments were conducted on the MUUFL Gulfport HSI and LiDAR data set and a remotely sensed soybean and weed data set. Results show improved, consistent performance on scene understanding and agricultural applications when compared to traditional fusion methods.

Urban environments pose a significant challenge for autonomous vehicles (AVs) as they must safely navigate while in close proximity to many pedestrians. It is crucial for the AV to correctly understand and predict the future trajectories of pedestrians to avoid collision and plan a safe path. Deep neural networks (DNNs) have shown promising results in accurately predicting pedestrian trajectories, relying on large amounts of annotated real-world data to learn pedestrian behavior. However, collecting and annotating these large real-world pedestrian datasets is costly in both time and labor. This paper describes a novel method using a stochastic sampling-based simulation to train DNNs for pedestrian trajectory prediction with social interaction. Our novel simulation method can generate vast amounts of automatically-annotated, realistic, and naturalistic synthetic pedestrian trajectories based on small amounts of real annotation. We then use such synthetic trajectories to train an off-the-shelf state-of-the-art deep learning approach Social GAN (Generative Adversarial Network) to perform pedestrian trajectory prediction. Our proposed architecture, trained only using synthetic trajectories, achieves better prediction results compared to those trained on human-annotated real-world data using the same network. Our work demonstrates the effectiveness and potential of using simulation as a substitution for human annotation efforts to train high-performing prediction algorithms such as the DNNs.

In classifier (or regression) fusion, the aim is to combine the outputs of several algorithms to boost overall performance. Standard supervised fusion algorithms often require accurate and precise training labels. However, accurate labels may be difficult to obtain in many remote sensing applications. This paper proposes novel classification and regression fusion models that can be trained given ambiguously and imprecisely labeled training data in which the training labels are associated with sets of data points (i.e., “bags”) instead of individual data points (i.e., “instances”) following a multiple-instance learning framework. Experiments were conducted based on the proposed algorithms on both synthetic data and applications such as target detection and crop yield prediction given remote sensing data. The proposed algorithms show effective classification and regression performance.

The Multiple Instance Choquet integral (MICI) for classifier fusion and an evolutionary algorithm for parameter estimation is presented. The Choquet integral has a long history of providing an effective framework for non-linear fusion. However, previous methods to learn an appropriate measure for the Choquet integral required accurate and precise training labels. In many applications, data-point specific labels are unavailable and infeasible to obtain. The proposed MICI algorithm allows for training with uncertain labels in which class labels are provided for sets of data points (i.e., “bags”) instead of individual data points (i.e., “instances”). The proposed algorithm is able to fuse multiple two-class classifier outputs by learning a monotonic and normalized fuzzy measure from uncertain training labels using an evolutionary algorithm. It produces enhanced classification performance by computing Choquet integral with the learned fuzzy measure. Results on both simulated and real hyperspectral data are presented in the paper.