Precipitation nowcasting is an important task, which can serve numerous applications such as urban alert and transportation. Previous studies leverage convolutional recurrent neural networks (RNNs) to address the problem. However, they all suffer from two inherent drawbacks of the convolutional RNN, namely, the lack of a memory cell to preserve the fine-grained spatial appearances and the position misalignment issue when combining current observations with previous hidden states. In this article, we aim to overcome the defects. Specifically, we propose a novel pseudo flow spatiotemporal LSTM unit (PFST-LSTM), where a spatial memory cell and a position alignment module are developed and embedded in the structure of LSTM. Upon the PFST-LSTM units, we develop a new sequence-to-sequence architecture for precipitation nowcasting, which can effectively combine the spatial appearances and motion information. Extensive empirical evaluations are conducted on synthetic MovingMNIST++ and CIKM AnalytiCup 2017 datasets. Our experimental results demonstrate the superiority of the proposed PFST-LSTM over the state-of-the-art competitors. To reproduce the results, we release the source code at: https://github.com/luochuyao/PFST-LSTM.

Equipping a deep model the ability of few-shot learning (FSL) is a core challenge for artificial intelligence. Gradient-based meta-learning effectively addresses the challenge by learning how to learn novel tasks. Its key idea is learning a deep model in a bi-level optimization manner, where the outer-loop process learns a shared gradient descent algorithm (called meta-optimizer), while the inner-loop process leverages it to optimize a task-specific base learner with few examples. Although these methods have shown superior performance on FSL, the outer-loop process requires calculating second-order derivatives along the inner-loop path, which imposes considerable memory burdens and the risk of vanishing gradients. This degrades meta-learning performance. Inspired by recent diffusion models, we find that the inner-loop gradient descent process can be viewed as a reverse process (i.e., denoising) of diffusion where the target of denoising is the weight of base learner but origin data. Based on this fact, we propose to model the gradient descent algorithm as a diffusion model and then present a novel conditional diffusion-based meta-learning, called MetaDiff, that effectively models the optimization process of base learner weights from Gaussian initialization to target weights in a denoising manner. Thanks to the training efficiency of diffusion models, our MetaDiff does not need to differentiate through the inner-loop path such that the memory burdens and the risk of vanishing gradients can be effectively alleviated for improving FSL. Experimental results show that our MetaDiff outperforms state-of-the-art gradient-based meta-learning family on FSL tasks.

The task of precipitation nowcasting is significant in the operational weather forecast. The radar echo map extrapolation plays a vital role in this task. Recently, deep learning techniques such as Convolutional Recurrent Neural Network (ConvRNN) models have been designed to solve the task. These models, albeit performing much better than conventional optical flow based approaches, suffer from a common problem of underestimating the high echo value parts. The drawback is fatal to precipitation nowcasting, as the parts often lead to heavy rains that may cause natural disasters. In this paper, we propose a novel interaction dual attention long short-term memory (IDA-LSTM) model to address the drawback. In the method, an interaction framework is developed for the ConvRNN unit to fully exploit the short-term context information by constructing a serial of coupled convolutions on the input and hidden states. Moreover, a dual attention mechanism on channels and positions is developed to recall the forgotten information in the long term. Comprehensive experiments have been conducted on CIKM AnalytiCup 2017 data sets, and the results show the effectiveness of the IDA-LSTM in addressing the underestimation drawback. The extrapolation performance of IDA-LSTM is superior to that of the state-of-the-art methods.

Remote Sensing Image Captioning (RSIC) presents unique challenges and plays a critical role in applications such as environmental monitoring, urban planning, and disaster management. Traditional RSIC methods often struggle to produce rich and diverse descriptions. Recently, with significant advancements in Vision-Language Models (VLMs), efforts have emerged to integrate these models into the remote sensing domain and to introduce richly descriptive datasets specifically designed to enhance VLM training. However, most current RSIC models generally apply only fine-tuning to these datasets without developing models tailored to the unique characteristics of remote sensing imagery. This paper proposes RS-MoE, the first Mixture of Expert based VLM specifically customized for remote sensing domain. Unlike traditional MoE models, the core of RS-MoE is the MoE Block, which incorporates a novel Instruction Router and multiple lightweight Large Language Models (LLMs) as expert models. The Instruction Router is designed to generate specific prompts tailored for each corresponding LLM, guiding them to focus on distinct aspects of the RSIC task. This design not only allows each expert LLM to concentrate on a specific subset of the task, thereby enhancing the specificity and accuracy of the generated captions, but also improves the scalability of the model by facilitating parallel processing of sub-tasks. Additionally, we present a two-stage training strategy for tuning our RS-MoE model to prevent performance degradation due to sparsity. We fine-tuned our model on the RSICap dataset using our proposed training strategy. Experimental results on the RSICap dataset, along with evaluations on other traditional datasets where no additional fine-tuning was applied, demonstrate that our model achieves state-of-the-art performance in generating precise and contextually relevant captions. Notably, our RS-MoE-1B variant achieves performance comparable to 13B VLMs, demonstrating the efficiency of our model design. Moreover, our model demonstrates promising generalization capabilities by consistently achieving state-of-the-art performance on the Remote Sensing Visual Question Answering (RSVQA) task.

Cross-view geo-localization is a critical task in various applications, such as smart city management and disaster monitoring. Current methods typically divide a satellite image into patches and use these patches to identify the geographic location of a query image. However, these methods can only provide the location of an image rather than the location of a specific object of interest. This makes it difficult to link these methods to GeoDatabases to obtain detailed information about a target object, such as its name and construction time. To overcome this limitation, we propose a novel problem of cross-view object geo-localization in a local region with high-resolution satellite images. This problem includes two main challenges: accurately identifying the location of an object and distinguishing the target object from others in satellite images. To address these challenges, we present a new Detection-based Geo-localization method called DetGeo, which consists of an object detection-based framework with a two-branch encoder and a query-aware cross-view fusion module. DetGeo uses cross-view images as input to the detector to provide object-level geo-localization. The fusion module employs cross-view spatial attention to focus on relevant areas of target objects during cross-view feature fusion. To evaluate our method, we constructed a new Cross-View Object Geo-Localization dataset called CVOGL, which comprises ground-view or drone-view images as query images and satellite-view images as geo-tagged reference images. Comprehensive experiments are conducted to demonstrate the effectiveness of our method on CVOGL. https://github.com/sunyuxi/DetGeo.