Despite growing interest in Audio-Visual Question Answering (AVQA), existing datasets often suffer from limited diversity, rigid formats, and insufficient integration of audio and visual modalities. To address these limitations, we introduce Valor32k-AVQA v2.0, a large-scale dataset containing 28,863 real-world videos and over 225,000 QA pairs, designed to support diverse and realistic multimodal understanding. The dataset features both open-ended and multiple-choice questions, each annotated with the required modality ( visual, audio, or audio-visual ) and question category ( description, action, count, temporal, location, or relative position ). All annotations-including questions, answers, and metadata-are generated through a fully automated prompting pipeline using GPT-4o, with human validation performed on a representative sample to ensure quality. We benchmark a few state-of-the-art models, with additional evaluations available on the project page, and observe that incorporating audio consistently improves performance during fine-tuning without compromising visual reasoning capabilities. These findings highlight that the audio signals in our dataset are not only well integrated, but also informative and complementary, establishing Valor32k-AVQA v2.0 as a valuable resource for developing and evaluating robust audio-visual question answering systems.

Ocean fronts have an important influence on global ocean–atmosphere interactions and marine fishery. Hence, it is of great significance to obtain the positions of the ocean fronts. However, current ocean front detection research confronts two challenges: scarcity of labeled data and limitations of ocean front detection algorithms. To address these two problems, we have collected and labeled an ocean front data set and proposed a new deep learning model for ocean front detection. For concreteness, due to the weak edge property of the ocean fronts, we formulate ocean front detection as a weak edge identification problem and propose the weak edge identification network (WEIN) for ocean front detection. WEIN consists of four convolutional blocks. Each block has a side output layer used to detect front edges at a specific image representation level. The side outputs are then fused to predict (detect) the locations of the ocean fronts. In this work, we adopt two metrics to measure the experimental results, i.e., the$F_{1}$-score and intersection over union (IoU). The experimental results with comparison to traditional and deep learning approaches demonstrate the superiority of WEIN for ocean front detection.

In this article, we propose a new semi-supervised method to detect changes occurring in a geographical area after a major event such as war, an earthquake or flood. The detection is made by processing a pair of bi-temporal remotely sensed images of the area under consideration. The proposed method adopts a patch-based approach, where successive pairs of patches from the input images are compared using a deep machine learning method trained with augmented data. Our main contribution consists of proposing an approach for generating a training dataset from unlabeled pair of input images. The genuine training patch-pairs are directly generated from the transformed maps of the image taken before the event, while the impostor patch-pairs are generated by pairing the image taken before the event with any images, from the Internet, with textures that resemble the change shown in the image taken after the event. Several experiments were conducted on pairs of images related to five major events. The obtained subjective results demonstrate the effectiveness of the proposed method.

The field of neuroevolution has received great attention in recent years due to its promising capability for developing well-performing models. It has been applied to many real-world problems ranging from medical diagnosis to autonomous robots. The choice of the evolutionary algorithm (EA) has a huge impact on the neuroevolution overall performance. Despite recent progress in the field, it is not clear what the best choice of EA is. The problem becomes more severe considering a dozen of EAs available for neuroevolution applications. In this paper, six state of the art EAs are applied for the task of autonomous robot navigation. These EAs are MultiVerse optimizer (MVO), moth-flame optimization (MFO), particle swarm optimization (PSO), cuckoo search (CS), Grey wolf optimizer (GWO) and bat algorithm. MLP networks are trained using these six evolutionary algorithms to solve the classification task related to the autonomous robot navigation. Comprehensive experiments are conducted using three datasets and obtained results are visually and statistically compared. To the best knowledge of the authors, comparison among the aforementioned algorithms has not been considered in the literature. It is found that neuroevolution methods perform well for the task of autonomous robot navigation. Amongst investigated EAs, MVOtrained achieves the highest and most consistent performance metrics.

In this letter, a novel decolorization method is proposed to convert color images into grayscale. The proposed method, called CorrC2G, estimates the three global linear weighting parameters of the color to gray conversion by correlation. These parameters are estimated directly from the correlations between each channel of the RGB image and a contrast image. The proposed method works directly on the RGB channels; it does not use any edge information nor any optimization or training. The objective and subjective experimental results on three available benchmark datasets of color to gray conversion, e.g., Cadik, CSDD, and Color250, show that the proposed decolorization method is highly efficient and comparable to recent state-of-the-art decolorization methods. The MATLAB source code of the proposed method is available at: http://www.synchromedia.ca/system/files/CorrC2G.m.

In this letter, a very simple no-reference image quality assessment (NR-IQA) model for JPEG compressed images is proposed. The proposed metric called median of unique gradients (MUG) is based on the very simple facts of unique gradient magnitudes of JPEG compressed images. MUG is a parameterless metric and does not need training. Unlike other NR-IQAs, MUG is independent to block size and cropping. A more stable index called MUG$^+$is also introduced. The experimental results on six benchmark datasets of natural images and a benchmark dataset of synthetic images show that MUG is comparable to the state-of-the-art indices in the literature. In addition, its performance remains unchanged for the case of the cropped images in which block boundaries are not known. The MATLAB source code of the proposed metrics is available athttps://dl.dropboxusercontent.com/u/74505502/MUG.mandhttps://dl.dropboxusercontent.com/u/74505502/MUGplus.m.

In this paper, we propose a new unsupervised change detection method designed to analyze multispectral remotely sensed image pairs. It is formulated as a segmentation problem to discriminate the changed class from the unchanged class in the difference images. The proposed method is in the category of the committee machine learning model that utilizes an ensemble of classifiers (i.e., the set of segmentation results obtained by several thresholding methods) with a dynamic structure type. More specifically, in order to obtain the final “change/no-change” output, the responses of several classifiers are combined by means of a mechanism that involves the input data (the difference image) under an iterative Bayesian-Markovian framework. The proposed method is evaluated and compared to previously published results using satellite imagery.

The first competition on the MultiSpectral Text Extraction (MS-TEx) from historical document images has been organized in conjunction with the ICDAR 2015 conference. The goal of this contest is evaluation of the most recent advances in text extraction from historical document images captured by a multispectral imaging system. The MS-TEx 2015 dataset contains 10 handwritten and machine-printed historical document images along with eight spectral images for each image. This paper provides a report on the methodology and performance of the five submitted algorithms by various research groups across the world. The objective evaluation and ranking was performed by using well-known evaluation metrics of binarization and classification.

This paper presents a novel preprocessing method of color-to-gray document image conversion. In contrast to the conventional methods designed for natural images that aim to preserve the contrast between different classes in the converted gray image, the proposed conversion method reduces as much as possible the contrast (i.e., intensity variance) within the text class. It is based on learning a linear filter from a predefined data set of text and background pixels that: 1) when applied to background pixels, minimizes the output response and 2) when applied to text pixels, maximizes the output response, while minimizing the intensity variance within the text class. Our proposed method (called learning-based color-to-gray) is conceived to be used as preprocessing for document image binarization. A data set of 46 historical document images is created and used to evaluate subjectively and objectively the proposed method. The method demonstrates drastically its effectiveness and impact on the performance of state-of-the-art binarization methods. Four other Web-based image data sets are created to evaluate the scalability of the proposed method.

This article deals with a serious form of degradation that often affects the readability of historical document images: the invisibility of text or ink. Due to wear over long periods of storage, the ink may become invisible to the human eye, an undesirable situation for scholars (i.e. Indian Ocean World project (IOW1, with whom we are working closely). Because only the class of ink is known a priori (reference), it can be considered as a target to be detected. This can be achieved by designing a linear filter that maximizes an energy function while minimizing the false detection of document image background elements. For each document image in which the ink is targeted, an internal reference is defined by a new self-referencing strategy. The proposed method is compared with a state-of-the-art methods, and validated on samples of real historical document images.

Human ground-truthing is the manual labelling of samples (pixels for example) to generate reference data without any automatic algorithm help. Although a manual ground-truth is more accurate than a machine ground-truth, it still suffers from mislabeling and/or judgement errors. In this paper we propose a new method of ground-truth estimation using multispectral (MS) imaging representation space for the sake of document image binarization. Starting from the initial manual ground-truth, the proposed classification method aims to select automatically some samples with correct labels (well-labeled pixels) from each class for the training phase, then reassign new labels to the document image pixels. The classification scheme is based on the cooperation of multiple classifiers under some constraints. A real data set of MS historical document images and their ground-truth is created to demonstrate the effectiveness of the proposed method of ground-truth estimation.

The extraction and analysis of useful information from old document images is very important into cultural heritage preservation. In advanced research, where the goal is to separate the foreground (in general, text) from the background, image restoration and pattern classification techniques are used. Most of these methods consist of classifying the pixels based on their gray-scale value. In this paper, we propose to perform foreground pattern extraction using regions-of-interest (ROI) analysis and a maximum likelihood classifier designed for multispectral document images. As contribution, a new feature vector is proposed to improve discrimination between patterns that is embedded in a simple statistical classification method. The results, which are promising, are compared to the state-of-the-art.

In this paper, a new unsupervised hierarchical approach to textured color images segmentation is proposed. To this end, we have designed a two-step procedure based on a grey-scale Markovian over-segmentation step, followed by a Markovian graph-based clustering algorithm, using a decreasing merging threshold schedule, which aims at progressively merging neighboring regions with similar textural features. This hierarchical segmentation method, using two levels of representation, has been successfully applied on the Berkeley Segmentation Dataset and Benchmark (BSDB, Martin et al., 2001). The experiments reported in this paper demonstrate that the proposed method is efficient in terms of visual evaluation and quantitative performance measures and performs well compared to the best existing state-of-the-art segmentation methods recently proposed in the literature.

The non-local means filter is one of powerful denoising methods which allows participation of far, but proper pixels in the denoising process. Although the weights of non-similar pixels are very small, high number of these pixels results in introduction of blur. In this work, we introduce an automatic and robust method to select the best candidate pixels based on their similarity to the target pixel. This method is based on graphs partitioning and uses Markovian clustering on the pixel adjacency graph (PAG). In this way, a set of relevant pixels is obtained that is used in weighted averaging for denoising each pixel. To evaluate the method, denoising of the natural images is conducted, and the results are compared to the standard NLM filter and the SVD-based method. The results are promising.