Explainable AI for Deep Visual Recognition: Evaluation, Methods, and Open Challenges

Explainable Deepfake Detection Challenge

Deepfake detection is moving beyond binary classification decisions toward systems that can also explain the visual evidence supporting those decisions. This transition is important for real-world verification settings, where diverse users need to understand not only whether an image is manipulated, but also why it is considered suspicious. The Explainable Deepfake Detection Challenge at ACM Multimedia 2026 is designed to benchmark this joint capability. Built on XPlainVerse, a million-scale benchmark for explainable deepfake detection, the challenge evaluates methods on image classification and grounded natural-language explanation generation. Participants submit a real/fake label together with two explanations for each image: a detailed complex explanation for technical users and a concise simple explanation for general users. The evaluation combines classification metrics with semantic similarity, simplicity, and intent-aware grounding metrics that assess whether explanations identify the relevant manipulated entities and supporting visual evidence. The methodologies developed through the challenge will contribute to the development of next-generation explainable deepfake detectors. Evaluation script, baseline models, and accompanying code are available on https://github.com/Abhijeet8901/XPlainVerse-ACMChallenge.

A Comprehensive Review of Deepfake Detection Techniques: From CNN-Based Models to Explainable Multimodal LLM Frameworks

With the progress of generative adversarial networks (GANs), diffusion models, and large-scale deep learning models, deepfake technology has developed rather quickly and becomes a significant challenge to digital integrity, cybersecurity, journalism, and forensic investigations. While conventional CNN-based deepfake detectors demonstrated high accuracy in detecting visual artifacts; it only fulfills the lack of robustness as well as cross domain generalization and interpretability. The review provides an in-depth survey of fifteen state-of-art methodologies including classical CNN models, temporal-spatial video recognition, transformer-based networks, explainable AI (XAI) models, and models that combine multimodal large language model (LLM) products. The taxonomy classifies the current techniques in terms of representational features, learning models, interpretability procedures and deployment capability. Particular attention is paid to explainable and human-friendly AI systems including Grad-CAM visualizations, captioning systems, and narrative generation modules. Some of the important issues include adversarial robustness, bias in the data sets, computational complexity, cross-dataset generalization, and computational limitations in the real world. Lastly, the release of new research directions in terms of creating reliable, scalable, and understandable deepfake detectors is described.

Can Vision-Language Models Reason about AI Edits in Images?

Detection and localization of AI-tampered images are critical for trustworthy AI, yet modern generative models have made such manipulations increasingly difficult to identify. While traditional binary classifiers can detect image tampering, they lack interpretability and generalization. Vision-Language Models (VLMs) offer a promising alternative due to their strong visual understanding and reasoning capabilities; however, existing approaches typically rely on supervised finetuning with curated explanations rather than exploiting their inherent reasoning capabilities. In this work, we investigate whether VLMs can be trained to reason about AI-generated image edits using reinforcement learning (RL) rather than explicit reasoning supervision. Motivated by the success in Group Relative Policy Optimization (GRPO), an RL technique that incentivizes the model to reason by generating thinking traces prior to giving the final answer, we propose a GRPO-based training framework that utilizes simple accuracy and format rewards. Given an input image, the model produces a structured reasoning trace and predicts whether the image has been tampered with. A lightweight segmentation model is then guided by the reasoning output to generate pixel-level localization masks. Experiments across multiple image manipulation datasets demonstrate that our approach achieves competitive detection and localization performance compared to state-of-the-art image forgery detectors, despite requiring substantially weaker supervision. We introduce effective intersection over union (eff-IoU), a unified metric to jointly evaluate detection and localization. These results suggest that reinforcement learning provides an effective and scalable mechanism for teaching VLMs to reason about AI-generated content.

Computer Vision in the Deep Learning Era: A Survey of Architectures, Datasets, and Applications

In the deep learning era, computer vision has advanced quickly, radically changing how machines interpret and evaluate visual data. Data-driven deep neural networks that can immediately learn complicated representations from images have essentially replaced traditional computer vision algorithms that depended on manually created feature extraction. With an emphasis on important architectures including Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and new hybrid models, this review article offers a thorough overview of contemporary deep learning-based computer vision. It also looks at popular benchmark datasets, such as ImageNet, COCO, and CIFAR, emphasizing how important they are for model building and assessment. The study also examines a variety of practical uses, from autonomous systems and facial recognition technology to object identification and medical picture analysis. Despite impressive advancements, a number of issues still exist, including as high processing demands, reliance on data, interpretability of models, and ethical issues with prejudice and privacy. The report concludes by outlining future research possibilities, with a focus on real-time deployment, multimodal learning, and efficient model construction. The purpose of this review is to offer an organized and perceptive overview of computer vision's present situation and promise in the deep learning age.

Vision Pretraining for Dense Spatial Perception

Dense spatial perception is essential for physical intelligence, where visual systems are expected to recover structured, metric, and actionable representations from pixel observations. Modern visual foundation models tend to prioritize semantic invariance, often at the expense of detailed spatial understanding. In this work, we study vision pretraining through a boundary-centric lens, motivated by the premise that boundaries and shape discontinuities offer essential cues for perceiving geometric properties. Concretely, we propose masked boundary modeling, a self-supervised paradigm that dynamically learns sub-pixel boundary representations and subsequently leverages the discovered boundary-bearing tokens as masked targets to facilitate dense visual token learning. By scaling this framework, we develop LingBot-Vision and demonstrate its efficacy across a diverse set of downstream vision tasks with DINOv3 as a strong baseline. Remarkably, LingBot-Vision drives the progression from LingBot-Depth 1.0 to LingBot-Depth 2.0 for depth completion, and thereby yields enhanced depth estimation, a key pillar for embodied artificial intelligence. Our findings reveal that boundary modeling goes beyond simple line segments and instead serves as a scalable pretraining principle for learning spatially structured visual representations.

Robustness evaluation of explainable AI methods under natural image corruptions