Grad-CAM++
PulseAugur coverage of Grad-CAM++ — every cluster mentioning Grad-CAM++ across labs, papers, and developer communities, ranked by signal.
6 day(s) with sentiment data
-
New deep learning benchmark for dental radiograph classification released
Researchers have developed a new deep learning benchmark for classifying periapical radiographs, addressing issues of patient-level data splits and cross-center validation. The benchmark, applied to the DentIRO dataset,…
-
Study reveals misalignment in explainable AI evaluation methods
A new study published on arXiv investigates the alignment of different evaluation strategies for explainable AI (XAI). Researchers compared subjective measures like trust and satisfaction, objective metrics such as task…
-
New lightweight AI model AgroVisNet aids crop disease classification
Researchers have developed AgroVisNet, a lightweight convolutional neural network designed for plant disease classification on devices with limited connectivity and computational power. This model, along with the expert…
-
New framework quantifies temporal explainability in medical AI video analysis
Researchers have developed a new quantitative framework to evaluate the temporal explainability of deep learning models used in echocardiographic video segmentation. This framework uses four metrics to assess temporal c…
-
New AI framework enhances glaucoma detection using attention and ensemble learning
Researchers have developed a novel framework for detecting glaucoma by combining attention-enhanced deep feature extraction with heterogeneous ensemble learning. This approach utilizes InceptionV3 and the Convolutional …
-
AI models achieve high accuracy in retinal disease classification and vessel segmentation
Researchers have developed a novel two-pipeline framework for analyzing retinal fundus images, combining disease classification with blood vessel segmentation. The framework fine-tuned eight ImageNet-pretrained CNNs for…
-
New method tackles rotation-induced drift in AI model explanations
Researchers have identified a significant issue with post-hoc saliency maps, such as Grad-CAM, used to audit AI model decisions. These maps exhibit a 'drift' when input images are rotated, even if the model's prediction…
-
New HiLRP method offers unified explanation for diverse Vision Transformers
Researchers have developed a new attribution method called HiLRP designed to provide a single, trustworthy explanation for Vision Transformer (ViT) models. Existing methods struggle with the diverse architectures of ViT…
-
Vision foundation models show promise for explainable diabetic retinopathy classification
Researchers have developed an explainable framework for classifying diabetic retinopathy (DR) using vision foundation models. The study evaluated DINOv2, CLIP, and Vision Transformer backbones with various transfer lear…
-
New LUX architecture enhances explainable endoscopic image captioning
Researchers have developed LUX, a novel graph-conditioned vision-language architecture designed for explainable endoscopic image captioning. This system addresses the limitations of current deep learning models by const…
-
Knowledge distillation can improve CNNs in data-scarce settings
Researchers have investigated knowledge distillation (KD) for training smaller, more efficient Convolutional Neural Networks (CNNs) by transferring knowledge from larger teacher models. While typically applied at the fi…
-
New HiRA-CAM method enhances CNN explainability
Researchers have introduced HiRA-CAM, a novel method for improving the explainability of convolutional neural networks (CNNs). This new technique builds upon the existing LayerCAM approach by adaptively utilizing activa…
-
AI framework improves appendicitis grading from ultrasound images
Researchers have developed AppendiGrade, a deep learning framework designed to improve the grading of appendicitis from ultrasound images. The system utilizes four pre-trained models, including InceptionV3, which achiev…
-
MagViT transformer framework enhances breast cancer detection accuracy
Researchers have developed MagViT, a novel interpretable multi-magnification transformer framework designed for breast histopathology classification. This model utilizes a ViT backbone to process images at four differen…
-
New AI method for clinical data explanations shows promise but struggles with real-world localization
Researchers have developed a novel method called "Pathology Transport" that utilizes optimal transport to create explanations for clinical AI models. This approach models the distributions of healthy and diseased patien…
-
AI model enhances diabetic retinopathy grading with uncertainty awareness
Researchers have developed a new pipeline for automated diabetic retinopathy (DR) grading that incorporates lesion-aware preprocessing, ordinal predictions, and uncertainty estimation. The system uses a specific feature…
-
AI radiology tool MIRROR separates findings from prose for auditable reports
Researchers have developed MIRROR, a prototype system designed to address issues in AI-driven radiology reporting. MIRROR separates the classification of findings from the generation of textual reports, ensuring that ge…
-
Explainable AI in Neurological Imaging: Methods Reviewed, Gaps Identified
A recent review of 77 studies on explainable artificial intelligence (XAI) in neurological medical imaging has identified key methods and significant gaps. The paper highlights techniques such as Grad-CAM and SHAP, whil…
-
2D CNNs improve plant trait retrieval from spectral images
Researchers have developed a new method for plant trait retrieval using hyperspectral spectroscopy by transforming 1D spectral data into 2D images. This approach, utilizing convolutional neural networks (CNNs) like Effi…
-
AmalthAI platform democratizes AI for cultural heritage analysis
A new open-source computer vision platform called AmalthAI has been developed to make AI tools more accessible to cultural heritage experts. The platform simplifies the process of dataset management, model training, and…