The problem the researchers set out to solve is fundamentally one of scarcity and visual difficulty. Deep learning detectors need large, well-annotated datasets, but collecting and labeling underwater imagery is labor-intensive and expensive. Existing resources such as Trash-ICRA19, with 5,720 images from marine remotely operated vehicle surveys, TrashCan 1.0 with 7,212 segmented images, UNO, CleanSea, and DeepPlastic all focus on marine, coastal, or deep-sea settings. Those environments differ substantially from shallow freshwater ponds in illumination, water clarity, object appearance, and background complexity, so models trained on them transfer poorly to pond conditions. The team also noted that prior freshwater-focused systems, such as camera-based floating plastic detectors and the YOLO-EcoFocus framework for inland waterways, target debris at or near the water surface rather than fully submerged waste on a pond floor.

PondDebris comprises 1,526 images and 3,224 bounding box annotations spread across two ecologically meaningful categories: biodegradable and non-biodegradable waste. Rather than simply sorting trash by material, the taxonomy reflects environmental persistence, a distinction that matters for automated waste management because biodegradable items such as fruit peels, vegetable matter, leaves, uncoated paper, and untreated wood will break down naturally within weeks to a few months, while plastic bottles, plastic bags, aluminum cans, polystyrene, and synthetic packaging will not. Under the annotators’ operational rule, treated or painted wood and coated paper counted as non-biodegradable because their surface treatments delay natural decomposition. The final tally included 1,537 biodegradable and 1,687 non-biodegradable instances.

The imagery was collected in a controlled laboratory aquarium designed to emulate the visual character of a shallow freshwater pond, complete with a sandy substrate, variable turbidity, changing illumination, water reflections, and partial occlusion of objects. A Crosstour CT8500 action camera recorded 4K UHD video across three sessions, the last under different artificial lighting to broaden visual diversity. Frames were extracted, screened to remove redundant or blurred shots, and resized to 512 by 512 pixels for training. The authors are explicit that this is a simulated proxy rather than an in-situ field dataset, a limitation they flag for future work, but the controlled setup allowed precise manipulation of debris placement and environmental variability while demonstrating that useful underwater benchmarks can be built with consumer-grade hardware.

Annotation quality received careful statistical treatment. Three annotators labeled the images in Roboflow, reviewed every annotation twice, and submitted a subset for independent verification by a faculty supervisor. On a stratified subsample re-annotated independently, class-level agreement measured by Cohen’s kappa reached 0.83, and bounding-box localization agreement, expressed as mean intersection-over-union, reached 0.87, both indicating substantial consistency. The dataset was split into 1,206 training, 183 validation, and 137 test images, and because the collection is modest in scale, the team supplemented the single split with five-fold cross-validation to guard against sampling variability. The dataset is released under a CC BY 4.0 license.

On the modeling side, UWAD-Net builds on YOLOv8s, the small variant of a single-stage detector that pairs a cross-stage-partial backbone with an anchor-free, decoupled head that separates classification from localization. YOLOv8s was chosen because it offered the best balance of accuracy and efficiency among six YOLO-family baselines evaluated, running at roughly 182 frames per second on an NVIDIA Tesla T4 GPU with about 11.17 million parameters. The key innovation is the integration of the Convolutional Block Attention Module, or CBAM, into the YOLOv8s architecture. CBAM applies channel attention and spatial attention in sequence: channel attention uses average and max pooling with a shared multilayer perceptron to learn which feature channels carry discriminative information, while spatial attention generates a two-dimensional map through a convolution that highlights informative regions and suppresses background clutter.

The placement of these attention modules was itself a design decision tested empirically. CBAM blocks were inserted after the P3, P4, and P5 feature pyramid levels produced by the YOLOv8s neck, immediately before the detection head, so that each scale receives refinement tuned to small, medium, and large objects respectively. In the ablation study, channel attention alone outperformed spatial attention alone, and inserting full CBAM at all three scales yielded the best overall configuration. The intuition is that underwater imagery suffers from wavelength-dependent light absorption, scattering, and turbidity, so channel attention can emphasize features that survive color degradation while spatial attention suppresses diffuse responses from suspended particles and substrate textures, sharpening the network’s focus on debris boundaries.

Trained for up to 150 epochs with the Adam optimizer, cosine learning-rate decay, and standard augmentations such as mosaic, horizontal flips, and color jitter, UWAD-Net achieved a mean average precision at IoU 0.50 of 0.7294, an mAP50-95 of 0.4517, precision of 0.8206, and recall of 0.6421 on the PondDebris test set, with five-fold cross-validation standard deviations of only a few thousandths, indicating stable performance across data partitions. The most consistent advantage over the YOLOv8s baseline was precision, which rose by 7.1 percentage points, meaning fewer false-positive detections, though the authors are candid that this came with a modest reduction in recall and slightly lower mAP, a precision-oriented trade-off rather than a uniform improvement. Against non-YOLO baselines, UWAD-Net outperformed Faster R-CNN with a ResNet-50-FPN backbone and the transformer-based RT-DETR on the same protocol, and the added attention machinery cost only a marginal increase in computation, keeping the model practical for resource-constrained edge deployment.

To probe generalization, the team fine-tuned a PondDebris-pretrained UWAD-Net on an external Underwater Debris Detection Dataset of 5,125 images spanning 15 litter categories, mapped onto the same two-class taxonomy. The transferred model reached an mAP50 of 0.6629 and precision of 0.7738 on the external test split, confirming that the learned underwater debris representations remain useful in a previously unseen environment, although recall dropped to 0.5724. The degradation reflects domain shift: the external data were captured at roughly 4 to 5 meters depth under medium turbidity, with irregularly shaped items such as wood and leaves proving harder to distinguish from the substrate than well-defined objects like bottles. The authors conclude that improving robustness will require larger, class-balanced datasets and domain-adaptation strategies, and that field-collected pond imagery is the natural next step for validating the framework in real freshwater ecosystems.

Subject of Research: Attention-enhanced deep learning for detecting biodegradable and non-biodegradable waste in freshwater pond environments

Article Title: UWAD-Net: attention-enhanced underwater waste detection in freshwater pond environments

Article References: S, B. B., Rathnasekaran, S. R., Patnaik, R. A., Anbalagan, S., & M, V. (2026). UWAD-Net: attention-enhanced underwater waste detection in freshwater pond environments. Frontiers in Water, 8, Article 1923979. https://doi.org/10.3389/frwa.2026.1923979

Image Credits: AI Generated

DOI: 10.3389/frwa.2026.1923979

Keywords: underwater waste detection, freshwater ponds, UWAD-Net, PondDebris dataset, CBAM attention, YOLOv8, object detection, computer vision, environmental monitoring, plastic pollution, transfer learning, deep learning

The central insight behind the architecture is that two very different kinds of evidence can reveal a forgery. The first is semantic: a spliced photograph often contains objects whose lighting, edges, or textures do not quite agree with their surroundings, and modern deep learning models have become remarkably good at picking up such visual inconsistencies. The second is statistical: when a digital image is saved as a JPEG file, the compression process leaves a characteristic pattern of errors that changes wherever a region has been pasted in, retouched, or re-saved. Error Level Analysis, or ELA, is a classic forensic technique that amplifies these discrepancies by measuring how much a picture degrades when it is recompressed at a known quality level. Regions that were manipulated after the original compression tend to show different error levels than untouched areas, making them glow under analysis.

ELA has long been valued by forensic practitioners for exactly this reason, but it comes with a practical drawback: computing it across large collections of images is computationally expensive, and interpreting its output by hand or with shallow classifiers is unreliable. The new study sidesteps this bottleneck by vectorizing the ELA computation so that it can run efficiently alongside the RGB stream, turning a slow manual step into a fast, parallelized preprocessing operation that feeds directly into the network. Rather than choosing between appearance and compression evidence, the authors treat both as first-class inputs, processing them in parallel from the very first layer of the model.

Concretely, the architecture consists of two lightweight encoder networks, one receiving the original RGB image and the other receiving its ELA map. Each encoder progressively extracts features at multiple scales, learning representations that capture both local texture anomalies and broader structural context. The two feature streams are then fused and passed into a single shared decoder, which reconstructs a pixel-level map indicating which regions of the image are likely to have been manipulated. Skip connections run from both encoders into the decoder, a design borrowed from the celebrated U-Net segmentation architecture, which preserves fine spatial detail that would otherwise be lost as features pass through successive downsampling stages. This matters enormously in forgery localization, where the difference between a useful forensic tool and a useless one often comes down to whether the output mask aligns precisely with the tampered pixels.

A second key ingredient is attention. The convolution blocks in both the encoder and decoder modules are augmented with the Convolutional Block Attention Module, or CBAM, a widely used mechanism that refines features along two complementary dimensions. The channel attention component learns which feature maps are most informative for distinguishing authentic content from manipulated content, effectively letting the network emphasize, say, the channels that respond to compression artifacts while suppressing those that encode irrelevant background detail. The spatial attention component then highlights the specific pixel locations where suspicious evidence is concentrated. Stacked together, these attention operations sharpen the discriminative power of the learned features without adding substantial computational cost, which is essential for a model intended to be practical rather than merely accurate on a benchmark.

The design choices reflect a deliberate focus on a particular family of forgeries: those that introduce compression inconsistencies. Image splicing, in which a fragment of one photograph is pasted into another, and retouching, in which parts of an image are altered or enhanced, both tend to disturb the uniform JPEG error signature of an authentic file. Because the pasted region typically originates from an image with a different compression history, its error level profile differs from that of its new surroundings. By feeding the ELA map directly into a dedicated encoder, the network can learn to exploit these discrepancies automatically, rather than relying on hand-crafted thresholds that break down when image quality, resolution, or compression settings vary.

This approach distinguishes the work from many recent deep learning methods for forgery detection, which have grown increasingly heavy. Survey literature in the field notes that while numerous proposed models achieve impressive results on standard benchmarks, many are either computationally intensive or generalize poorly to datasets they were not trained on. The authors position their model as a lighter alternative: by using lightweight encoders and an efficient vectorized ELA pipeline, the system aims to deliver competitive localization performance without the enormous parameter counts and inference costs of large vision transformers or heavily engineered multi-stage pipelines. In forensic practice, where investigators may need to triage thousands of images, that efficiency is not a luxury but a requirement.

The evaluation strategy follows the conventions of the field, drawing on established benchmark datasets for manipulated images, including collections such as CASIA, CoMoFoD, and related corpora that have become standard proving grounds for splicing and copy-move detection methods. The reported results show competitive performance on these benchmarks, indicating that the dual-stream design with CBAM-augmented blocks can hold its own against existing approaches while remaining computationally lean. The authors also tested the model on a large collection of real and fake faces hosted publicly on Kaggle, extending the evaluation beyond generic object scenes to the portrait imagery that dominates online misinformation.

Perhaps the most scientifically honest part of the study is what it admits. In cross-dataset evaluation—training on some datasets and testing on a completely unseen one—the model, like most of its predecessors, struggled to generalize to the full diversity of real-world manipulations. This is a well-documented weakness across the entire field of image forensics: models tend to learn dataset-specific cues, such as the compression settings or camera characteristics of a particular corpus, rather than universally valid traces of tampering. When those cues shift, performance drops. The authors explicitly flag this as an important avenue for future research, a candid acknowledgment that benchmark success does not yet translate into forensic reliability in the wild, where adversaries can re-encode, resize, and filter images precisely to erase the very artifacts detectors rely on.

The stakes for this line of research could hardly be higher. Generative tools now allow anyone to fabricate convincing photographs in seconds, and manipulated images have been documented influencing elections, financial markets, and public health debates. Passive forensic techniques—those that detect tampering without access to the original file or cryptographic signatures—are among the few remaining checks on this flood of synthetic and doctored content. By showing that a classical forensic signal like ELA can be vectorized, paired with deep visual features, and refined with attention mechanisms inside a compact encoder-decoder network, the study offers a template for building detectors that are simultaneously fast, accurate, and grounded in the physics of image formation. The road to robust, generalizable forensics remains long, but this work demonstrates that the most promising path may lie not in abandoning classical techniques for deep learning, but in teaching them to work together.

Subject of Research: Dual-stream CNN with RGB–ELA inputs and CBAM attention for detecting and localizing image forgeries

Article Title: An optimized dual-stream CNN with vectorized RGB–ELA inputs and CBAM attention for image forgery detection

Article References: Jaiswal, A. K., Nizami, A., & Srivastava, R. (2026). An optimized dual-stream CNN with vectorized RGB–ELA inputs and CBAM attention for image forgery detection. Neural Computing and Applications, 38(19), Article 796. https://doi.org/10.1007/s00521-026-12534-w

Image Credits: AI Generated

DOI: 10.1007/s00521-026-12534-w

Keywords: image forgery detection, digital image forensics, error level analysis, convolutional neural network, CBAM attention, feature fusion, image splicing, forgery localization, deep learning, U-Net, JPEG compression artifacts, computer vision

The research team, led by Ha Gia Son of the Vietnam Academy of Science and Technology, the Vietnam-Hungary Industrial University, and the Artificial Intelligence Research Center at VNU Information Technology Institute, together with Tran Manh Tuan of Thuyloi University and Le Hoang Son of VNU, focused on a challenge that matters most in regions where climate change is intensifying rainfall extremes. Northwest Vietnam, with its steep terrain and vulnerable mountain communities, served as one of the study’s testing grounds. In such settings, a forecast that is accurate in average pixel value but blurry in structure can still fail to warn residents about the localized downpour that actually triggers a flash flood.

At the heart of SR2-GAN is a generator built on a Convolutional Long Short-Term Memory network, or ConvLSTM, a recurrent architecture that treats radar frames as spatiotemporal tensors and learns how rainfall patterns evolve across both space and time. ConvLSTM networks have become a workhorse of radar echo extrapolation because they can, in principle, capture the motion, growth, and decay of precipitation cells. What the Vietnamese team added is a set of refined Convolutional Block Attention Modules, known as CBAM, which were originally introduced in computer vision research in 2018. These attention modules allow the network to weight which spatial regions and which feature channels matter most at each step of the prediction, helping the model concentrate on the dynamic features of rainfall rather than on static background clutter.

The second half of the adversarial pair is a discriminator that evaluates the generated radar frames at the patch level rather than judging each image as a whole. This approach, popularized by image-to-image translation frameworks such as pix2pix, asks whether small local windows of the forecast look realistic. In the context of precipitation nowcasting, patch-level discrimination pushes the generator to produce sharper edges around storm cells and more coherent internal structure, instead of the smeared, averaged-looking outputs that plague models trained purely on pixel-wise error minimization.

Perhaps the most consequential design decision in SR2-GAN is its composite multi-objective loss function. The training objective jointly combines three terms: the adversarial loss that rewards realism, the mean squared error (MSE) that penalizes pixel-level deviations from the true radar observations, and the structural similarity index measure (SSIM) that rewards preservation of morphological characteristics such as the shape, contrast, and texture of rain bands. Each term pulls the network in a slightly different direction. MSE alone tends to produce blurry forecasts because averaging over possible futures smears out uncertainty; adversarial loss alone can hallucinate plausible-looking but physically wrong structures. By balancing all three, the framework seeks forecasts that are simultaneously faithful in intensity and faithful in form.

The empirical results are striking. On radar data from the Pha Din station in Dien Bien, Vietnam, SR2-GAN reduced MSE by 44.4 percent compared with a CNN-GRU baseline, a hybrid of convolutional and gated recurrent layers that the same research community has previously applied to rainfall forecasting. On the DWD dataset from the German Weather Service, the reduction was 25 percent. Structural fidelity improved as well: SSIM rose by 12.8 percent on the Vietnamese data and 4.5 percent on the German data. The fact that the gains hold across two geographically and climatologically distinct datasets—one tropical and mountainous, one mid-latitude European—suggests the architecture is not simply overfitting to the quirks of a single radar site.

The comparisons did not stop at the CNN-GRU baseline. The authors report that SR2-GAN outperforms advanced generative competitors, including S2R-GAN, the team’s own earlier spatiotemporal GAN for radar image nowcasting presented at the ICTA 2025 conference, and Rad-cGAN, a conditional GAN for radar-based precipitation nowcasting developed for multiple dam domains and published in Geoscientific Model Development in 2022. The decisive advantages appear in maintaining spatiotemporal consistency across forecast steps and in delivering more stable long-range predictions—precisely the qualities that determine whether a nowcasting system can be trusted several steps into the future rather than just one or two frames ahead.

The significance of this work becomes clearer when placed against the broader arc of the field. Traditional numerical weather prediction models, such as prototypes of the Weather Research and Forecasting model and the physical parameterization schemes developed at ECMWF, excel at longer horizons but are computationally expensive and can struggle with the fine-grained, rapidly evolving convective systems that dominate short-term rainfall. Statistical approaches like SARIMA and support vector machines capture temporal patterns but largely ignore spatial structure. Deep learning has reshaped the landscape: PredRNN introduced spatiotemporal LSTM units for predictive learning, and in 2023 the NowcastNet system published in Nature demonstrated skillful nowcasting of extreme precipitation. SR2-GAN joins this lineage while tailoring the recipe to the practical constraints of radar-image sequence prediction, including in data-sparse developing regions.

For the researchers involved, the motivation is not abstract. The team has a history of building forecasting tools for the Pha Din radar station, including a 2024 model integrating CNN, gated recurrent units, and a genetic algorithm for rainfall forecasting from radar images. The progression from those hybrid statistical-machine-learning systems to a full generative adversarial framework reflects a deliberate bet that realism-aware training is the key to usable early warnings. In mountainous northern Vietnam, where the impacts of climate change are increasingly severe, an extra few minutes of accurate, spatially coherent warning can translate directly into evacuations completed before flash floods arrive.

The study also lowers barriers for other research groups. The German DWD dataset is publicly available through a Zenodo repository, the Pha Din radar data is published through a Mendeley Data repository, and the authors have released their code on GitHub. The research received no dedicated funding, and the authors declare no competing interests. As generative models continue their migration from image synthesis into the geosciences, SR2-GAN offers a concrete demonstration that attention-enhanced recurrent generators, disciplined by a carefully balanced composite loss, can turn raw radar reflectivity into sharper, more trustworthy glimpses of the very near future—and that such glimpses may soon be within reach of the communities that need them most.

Subject of Research: Deep learning-based generative adversarial networks for multi-step radar image precipitation nowcasting

Article Title: SR2-GAN: A Novel Generative Adversarial Network for Radar Image Sequence Nowcasting

Article References: Son, H. G., Tuan, T. M., & Son, L. H. (2026). SR2-GAN: A Novel Generative Adversarial Network for Radar Image Sequence Nowcasting. Mobile Networks and Applications. https://doi.org/10.1007/s11036-026-02532-6

Image Credits: AI Generated

DOI: 10.1007/s11036-026-02532-6

Keywords: SR2-GAN, generative adversarial network, radar nowcasting, precipitation forecasting, ConvLSTM, CBAM attention, SSIM, deep learning, spatiotemporal prediction, Vietnam, DWD dataset, early warning systems