In many visual systems, visual tracking often bases on RGB image sequences, in which some targets are invalid in low-light conditions, and tracking performance is thus affected significantly. Introducing other modalities such as depth and infrared data is an effective way to handle imaging limitations of individual sources, but multi-modal imaging platforms usually require elaborate designs and cannot be applied in many real-world applications at present. Near-infrared (NIR) imaging becomes an essential part of many surveillance cameras, whose imaging is switchable between RGB and NIR based on the light intensity. These two modalities are heterogeneous with very different visual properties and thus bring big challenges for visual tracking. However, existing works have not studied this challenging problem. In this work, we address the cross-modal object tracking problem and contribute a new video dataset, including 654 cross-modal image sequences with over 481K frames in total, and the average video length is more than 735 frames. To promote the research and development of cross-modal object tracking, we propose a new algorithm, which learns the modality-aware target representation to mitigate the appearance gap between RGB and NIR modalities in the tracking process. It is plug-and-play and could thus be flexibly embedded into different tracking frameworks. Extensive experiments on the dataset are conducted, and we demonstrate the effectiveness of the proposed algorithm in two representative tracking frameworks against 19 state-of-the-art tracking methods. Dataset, code, model and results are available at https://github.com/mmic-lcl/source-code.

Outlier-robust principal component analysis (ORPCA) has been broadly applied in scientific discovery in the last decades. In this paper, we study online ORPCA, an important variant that addresses the practical challenge that the data points arrive in a sequential manner and the goal is to recover the underlying subspace of the clean data with one pass of the data. Our main contribution is the first provable algorithm that enjoys comparable recovery guarantee to the best known batch algorithm, while significantly improving upon the state-of-the-art online ORPCA algorithms. The core technique is a robust version of the residual norm which, informally speaking, leverages not only the importance of a data point, but also how likely it behaves as an outlier.

Full-waveform satellite LiDAR can be used to retrieve the terrestrial surface information by decomposing its received waveforms. However, it is challenging to accurately extract the parameters of each component from a non-Gaussian overlapped waveform, which happens in steep mountain or urban areas. Therefore, a synthetic algorithm with the boosted Richardson–Lucy (RL) deconvolution, layered extraction, and gradient descent is proposed to implement the skew-normal decomposition for the received waveforms. To validate the performance of the proposed algorithm, we developed waveform decomposition experiments for three types of data, including known-parameter waveforms, simulated waveforms, and Global Ecosystem Dynamics Investigation (GEDI) satellite LiDAR waveforms. Meanwhile, we figured out the evaluation metrics involving correlation coefficients (CCs); root mean square errors (RMSEs); extracted parameter errors; and successful, missing, and unwanted rates for the decomposed waveforms. Through comparing the decomposed results of the proposed algorithm and the classical direct Gaussian decomposition (DGD) algorithms, we discovered that 1) the average CC has a growth of 4% and the average RMSE has a reduction of 60%; 2) the average errors of extracted amplitude, peak position, and pulsewidth have mitigated with 3.9%, 2.2%, and 5.1%, respectively; and 3) the successful detection rate increases by 40% and the unwanted and the missing rate decrease by 5% and 35% for the 2000 groups of known-parameter waveforms. In addition, the average CCs have slight growth of 3% and 1.2%, and the average RMSEs have significant reductions of 43% and 49% for the simulated and GEDI LiDAR waveforms, respectively. This research provides a preferable waveform decomposing approach conductive to characterizing the terrestrial information from the overlapping skew-normal full waveforms.

We consider online forecasting problems for non-convex machine learning models. Forecasting introduces several challenges such as (i) frequent updates are necessary to deal with concept drift issues since the dynamics of the environment change over time, and (ii) the state of the art models are non-convex models. We address these challenges with a novel regret framework. Standard regret measures commonly do not consider both dynamic environment and non-convex models. We introduce a local regret for non-convex models in a dynamic environment. We present an update rule incurring a cost, according to our proposed local regret, which is sublinear in time T. Our update uses time-smoothed gradients. Using a real-world dataset we show that our time-smoothed approach yields several benefits when compared with state-of-the-art competitors: results are more stable against new data; training is more robust to hyperparameter selection; and our approach is more computationally efficient than the alternatives.