Fréchet inception distance
PulseAugur coverage of Fréchet inception distance — every cluster mentioning Fréchet inception distance across labs, papers, and developer communities, ranked by signal.
9 day(s) with sentiment data
-
New Causal Variational Deep Embedding framework tackles confounded image generation
Researchers have introduced CauVaDE (Causal Variational Deep Embedding), a novel framework designed to address challenges in deep generative models that inherit spurious associations from training data due to unobserved…
-
New AI frameworks tackle unpaired image translation with advanced control
Researchers have developed two new frameworks for unpaired image-to-image translation, a task that involves altering an image's appearance while preserving its content without relying on paired examples. PRISM uses a di…
-
GSRAIN method synthesizes controllable rainfall for 3D driving scenes
Researchers have developed GSRAIN, a novel method for synthesizing realistic rainfall in 3D Gaussian Splatting (3DGS) driving scenes. This technique integrates a high-frequency raindrop model derived from real-world dat…
-
New diffusion model generates satellite imagery using wavelet-domain conditioning
Researchers have developed a novel diffusion framework that utilizes wavelet-domain conditioning for generating satellite imagery from cartographic data. This approach, built upon ControlNet and a Stable Diffusion backb…
-
Diffusion models' generation quality linked to data alignment and pseudorandom inputs
New research explores how the quality of generated images in diffusion models is affected by their internal mechanisms. One study identifies "expert-data alignment" as the key factor, suggesting that routing image gener…
-
DualDiT: Diffusion Transformer generates realistic OCT images and segmentation masks
Researchers have developed DualDiT, a novel conditional dual-output Diffusion Transformer designed for generating both optical coherence tomography (OCT) images and their corresponding segmentation masks. This approach …
-
New PARSE framework enhances concept erasure in diffusion models
Researchers have developed a new training-free framework called PARSE (Preservation-aware Adaptive Ranked Subspace Expansion) designed to improve concept erasure in text-to-image diffusion models. Existing methods often…
-
New watermarking technique for diffusion models preserves fidelity
Researchers have developed a new method called Latent Angular Watermarking (LAW) for embedding robust watermarks into diffusion models. This technique operates in the latent space, ensuring it doesn't interfere with the…
-
AI framework reduces dental implant artifacts in CBCT scans
Researchers have developed an unsupervised deep learning framework using a fine-tuned Cycle-Consistent Adversarial Network (CycleGAN) to reduce metal artifacts in dental cone-beam computed tomography (CBCT) scans. This …
-
Privacy-Preserving CT Slice Generation Framework Unveiled
Researchers from DS@GT ARC have developed a novel framework for generating synthetic lung CT slices that prioritizes privacy. Their approach integrates Optimal Transport Conditional Flow Matching with a post-generation …
-
New LLIFT framework generates realistic medical images for AI validation
Researchers have developed a new framework called Local Label-Informed Feature Transfer (LLIFT) to generate semi-synthetic brain MRI images with realistic lesions. This method aims to create more reliable ground-truth d…
-
AI generates substation meter defect images with limited data
A research paper proposes a novel framework to generate realistic defect images for substation meters, addressing the challenge of limited annotated samples. The method integrates Knowledge Embedding and Hypernetwork-Gu…
-
AI generates novel motifs for Indonesian Ulos textiles
Researchers have developed a generative AI framework to create novel motifs for traditional Ulos textiles, a cultural heritage of the Batak people in North Sumatra, Indonesia. By fine-tuning Protogen v3.4 and Stable Dif…
-
Quantum circuits show promise and challenges in AI generative models
Researchers are exploring the integration of quantum circuits into AI models, particularly for generative tasks like image synthesis and quantum circuit optimization. One study on quantum circuit synthesis found that wh…
-
ContiStain framework improves virtual IHC staining with MoE and relation-preserving distillation
Researchers have developed ContiStain, a novel framework designed to improve the performance of virtual immunohistochemistry (IHC) staining models when dealing with sequentially acquired data. This method utilizes a mix…
-
AI models show geographic bias in generated urban scenarios
A new research paper published on arXiv highlights significant geographic and diversity deficits in AI-generated urban scenarios. Researchers evaluated diffusion models like FLUX 1-schnell and Stable Diffusion 3.5 Large…
-
New SAGE method improves safety alignment in text-to-image models
A new research paper published on arXiv introduces StructureAware Geometric Regularization (SAGE), a novel method for improving the safety alignment of text-to-image diffusion models. Current alignment techniques often …
-
Momentum Guidance enhances flow-based image generation quality
Researchers have introduced Momentum Guidance (MG), a new technique designed to enhance the quality of images generated by flow-based models. MG works by extrapolating the current velocity along the ODE trajectory, impr…
-
ModaFlow framework enhances virtual try-on with modality-aware guidance
Researchers have developed ModaFlow, a novel framework for high-fidelity virtual try-on that improves garment semantic preservation and body geometry adaptation. The system utilizes a modality-aware guidance scheme, inc…
-
Review details Neural Architecture Search for Generative Adversarial Networks
This paper offers a comprehensive review of Neural Architecture Search (NAS) techniques applied to Generative Adversarial Networks (GANs). It categorizes and compares various NAS methods, focusing on search strategies, …