Efficiently processes high res images by selecting relevant patches.
problem High memory and compute requirements for processing large images.
method Differentiable Top-K operator to select relevant patches.
result End-to-end trainable model using backpropagation.
We have successfully implemented the "Learn to Pay Attention" model of attention mechanism in convolutional neural networks, and have replicated the results of the original paper in the categories of image classification and fine-grained recognition.
Paper proposes redundancy-free features for zero-shot object recognition.
problem Redundant visual features degrade zero-shot object recognition.
method Project original features into a new, statistically independent space.
result RFF-GZSL achieves competitive results on benchmark datasets.
LDA improves image classification accuracy with fewer features.
problem Fine-grained image classification with pretrained features.
method Supervised dimensionality reduction with LDA before linear probing.
result LDA improves accuracy over full features in 11 out of 12 configurations.
Tutorial on interpreting SPECT images for PD recognition using AI.
problem Difficulty in interpreting complex DL models for clinical use.
method Evaluation of six interpretation methods on four DCNN architectures.
result Guided backpropagation and SHAP methods are suitable for PD recognition.
A new method transcribes complex structured images like musical scores.
problem Transcribing content from images with complex internal structure.
method Hierarchical Spotlight Transcribing Network (STN) framework with two-stage approach.
result Demonstrated effectiveness through experiments on various structural image datasets.
Compact bilinear pooling approximates covariance features for faster training.
problem Efficiently approximating covariance features for faster training.
method Compact bilinear pooling extended to polynomial approximations of covariance features.
result The proposed method achieves comparable accuracy with fewer dimensions.
Proposes a new method to generate unrestricted adversarial examples.
problem Generating unrestricted adversarial examples without norm constraints.
method Leveraging state-of-the-art generative models to manipulate image fine-grained aspects.
result Our adversarial images look indistinguishable from natural images and can bypass certified defenses.
Paper reduces neural network complexity for image classification.
problem High computational complexity in deep neural networks.
method Proposes a two-step classification process: coarse-grain and fine-grain.
result Achieves similar accuracy with less computational complexity.
ECN framework improves training on noisy structured labels.
problem Structured errors in fine-grained annotations lead to biased models.
method Error-Correcting Networks (ECN) framework.
result ECN improves fine-grained annotation prediction.
Learning image representations to capture fine-grained semantics has been a challenging and important task enabling many applications such as image search and clustering. In this paper, we present Graph-Regularized Image Semantic Embedding (Graph-RISE), a large-scale neural graph learning framework that allows us to tr…
Selecting the most appropriate data examples to present a deep neural network (DNN) at different stages of training is an unsolved challenge. Though practitioners typically ignore this problem, a non-trivial data scheduling method may result in a significant improvement in both convergence and generalization performanc…
Pixel-wise relevance method shows how CNNs classify faces, varying across datasets and tasks.
problem Interpreting black-box CNN face recognition models.
method Layer-wise relevance propagation (LRP) applied to VGG-16 models trained for face recognition.
result Relevance maps are generally stable across random initializations and tasks, but less so across pretraining datasets.
Paper introduces rehearsal-free continual learning for small, non-i.i.d. batches in robotic vision.
problem Learning new objects and improving recognition in a changing robotic environment.
method Two rehearsal-free continual learning techniques (CWR* and AR1*) for small, non-i.i.d. batches.
result AR1* outperforms other techniques by more than 15% in some cases.
CLAREL improves zero-shot learning by using per-image semantic supervision and metric rescaling.
problem Fine-grained cross-modal representation learning for zero-shot classification.
method Instance-based deep metric learning in joint visual and textual space, using per-image semantic supervision and metric rescaling.
result CLAREL consistently outperforms existing approaches on fine-grained zero-shot learning datasets.
HW2MP-GAN tackles ancient handwritten text recognition.
problem Automatic text recognition from ancient handwritten records.
method Conditional Generative Adversarial Network (HW2MP-GAN) with Sliced Wasserstein distance and U-Net architectures.
result HW2MP-GAN outperforms state-of-the-art models in image-to-image translation and handwritten recognition.
DeepBDC improves few-shot classification by measuring joint distributions of image features.
problem Few-shot classification with limited training data.
method DeepBDC method using deep learning and Brownian Distance Covariance.
result DeepBDC significantly outperforms existing methods on various benchmarks.
Fawkes protects images from unauthorized facial recognition models.
problem Unauthorized training of facial recognition models poses privacy risks.
method Fawkes adds imperceptible pixel-level changes (cloaks) to images before release.
result Fawkes can protect images from misidentification by 95% and 80% even when clean images are leaked.
New text-to-image diffusion models improve scene understanding for AI agents.
problem Fine-grained scene understanding for AI agents from text and images.
method Pre-trained text-to-image diffusion models optimized for generating images from text prompts.
result Policies learned with Stable Control Representations outperform state-of-the-art approaches on various control tasks.
Improved object detection for scientific document images.
problem Current object detectors fail to accurately localize regions in scientific document images.
method Revised R-CNN model with region embedding for fine-grained proposals.
result 17% mAP improvement over standard object detection models.
AdarGCN tackles noisy web images in few-shot learning.
problem Few-shot learning with limited training samples from both source and target classes.
method AdarGCN combines LDN for web image denoising and GCN for FSL, using adaptive aggregation.
result AdarGCN outperforms conventional methods in FSFSL and conventional FSL settings.
L*ReLU improves deep learning for fine-grained image classification.
problem Fine-grained image classification requires specific AFs.
method Proposes L*ReLU, piece-wise linear AFs for deep learning.
result L*ReLU achieves superior results on FGVC datasets.
It is proposed a new code for contours of plane images. This code was applied for optical character recognition of printed and handwritten characters. One can apply it to recognition of any visual images.
Proposes a method for weakly-supervised object localization to improve few-shot learning.
problem Challenges of few-shot learning, especially with fine-grained categories.
method Introduces a Self-Attention Based Complementary Module (SAC Module) for weakly-supervised object localization.
result Significantly outperforms state-of-the-art methods on benchmark datasets, especially for fine-grained few-shot tasks.
Generates realistic faces from detailed textual descriptions.
problem Face generation from fine-grained textual descriptions.
method Conditional GAN model with DC-GAN and GAN-CLS loss, using CelebA dataset with generated captions.
result Promising results in generating diverse face images from fine-grained textual descriptions.
Improves naturalness in TTS samples using quantized VAE and auto-regressive prosody.
problem Discontinuous and unnatural speech from standard VAE priors.
method Discretized latent features using vector quantization (VQ), and separately trained autoregressive (AR) prior model.
result Significantly improves naturalness in random sample generation.
Deep Triplet Networks improve brain imaging modality recognition with limited data.
problem Efficiently recognizing new imaging modalities with scarce training data.
method Few-shot learning model based on Deep Triplet Networks.
result The model outperforms traditional CNN classifiers in modality recognition with limited data.
New approach improves human activity recognition with wearables.
problem Improving human activity recognition with wearables.
method Exploiting latent relationships between multi-channel sensor modalities, data-agnostic augmentation, and a classification loss criterion.
result Achieves new state-of-the-art performance on four diverse activity recognition benchmarks.
Paper introduces privacy-preserving few-shot learning for images.
problem Privacy risk in few-shot learning systems.
method Discrete embedding vectors and one-way hash functions.
result Achieves computational pan privacy without storing embeddings.
Proposes Moment Exchange to use moments in image recognition models, improving generalization.
problem Discarding moments in image recognition models reduces stability and training time.
method Moment Exchange: replaces moments of learned features with another image's moments and interpolates labels.
result Improves generalization of recognition models across multiple datasets.
AI enhances pollen recognition in veterinary imaging using holographic microscopy.
problem Challenges in recognizing pollen in holographic images due to speckle noise and artifacts.
method Training YOLOv8s and MobileNetV3L on dual-modality dataset, employing WGAN-SN for synthetic data augmentation.
result GAN-based augmentation improves object detection and classification in holographic images, closing the performance gap.
The paper addresses bias in visual recognition models by reweighting observations.
problem Bias in deep neural networks trained on biased image databases.
method Reweighting observations based on known biasing mechanisms to form a nearly debiased estimator.
result The approach can remedy representativeness issues in visual recognition systems.
Face recognition system trained with noisy labels.
problem Label noise in training deep learning classifiers.
method Review and apply recent methods to manage noisy annotations.
result Improved performance of face recognition system with noisy labels.
Efficient algorithms learn from coarse labels instead of fine grained ones.
problem Learning from coarse labels when fine labels are unavailable.
method Formalized coarse label settings, used a reduction for SQs, and provided efficient algorithms.
result Any problem learnable from fine labels can be learned efficiently from coarse labels.
DSS networks use scale-equivariant cross-correlations to improve image recognition.
problem Improving image recognition by exploiting scale invariance.
method Constructing scale-equivariant cross-correlations based on scale-spaces and semigroups.
result Demonstrated utility on Patch Camelyon and Cityscapes datasets.
Proposes a new metric learning method for image recognition.
problem Improving image recognition performance using learned distance representations.
method Introduces a Generalized Hybrid Metric Loss (GHM-Loss) to learn hybrid proximity features combining geometric and probabilistic spaces.
result Demonstrates superior performance compared to existing methods on public datasets.
ChemGrapher uses deep learning to automatically convert chemical compound images into accurate graphs.
problem Automatically converting chemical compound images into accurate graphs with correct bond multiplicity and stereochemical information.
method Developed a deep neural network model for optical compound recognition, including segmentation and classification models.
result Significant error reductions in bond multiplicity and stereochemical information compared to existing tools.
Traditional recognition methods typically require large, artificially-balanced training classes, while few-shot learning methods are tested on artificially small ones. In contrast to both extremes, real world recognition problems exhibit heavy-tailed class distributions, with cluttered scenes and a mix of coarse and fi…
ViewFool identifies adversarial viewpoints to test image recognition robustness.
problem Lack of robustness to viewpoint changes in visual recognition models.
method Neural Radiance Fields (NeRF) and entropic regularizer to find adversarial viewpoints.
result Common image classifiers are highly vulnerable to generated adversarial viewpoints.
Image recognition using Deep Learning has been evolved for decades though advances in the field through different settings is still a challenge. In this paper, we present our findings in searching for better image classifiers in offline and online environments. We resort to Convolutional Neural Network and its variatio…
Enhances 2D face recognition with 3D features using active illumination.
problem Improving robustness of 2D face recognition to spoofing attacks and low-light conditions.
method Projecting a high spatial frequency pattern onto the face to recover 3D information and a 2D image simultaneously.
result Significantly boosts face recognition performance and dramatically improves robustness to spoofing attacks.
Transforms improve CNNs' invariance to image transformations.
problem Current CNN models lack robustness to spatial transformations.
method Randomly transform feature maps during training to learn invariant representations.
result Significant improvements on benchmark tasks, including image recognition and retrieval.
Simple noise training improves image recognition robustness.
problem Modern neural networks struggle with unseen image corruptions.
method Additive Gaussian and Speckle noise training, adversarial training against worst-case noise.
result Significant improvement in robustness on ImageNet-C and MNIST-C benchmarks.
Two-stage neural network for few-shot image recognition.
problem Few-shot image recognition for novel categories.
method Multi-layer neural network with feature extraction and classification stages.
result Competitive performance on four standard datasets.
Thanks to recent advances in deep neural networks (DNNs), face recognition systems have become highly accurate in classifying a large number of face images. However, recent studies have found that DNNs could be vulnerable to adversarial examples, raising concerns about the robustness of such systems. Adversarial exampl…
Predict steering angles of self-driving cars from images.
problem Predicting steering angles for self-driving cars using image data.
method Used deep learning techniques like Transfer Learning, 3D CNN, LSTM, and ResNet to predict steering angles.
result Both models placed in the top ten of Udacity's challenge.
This paper examines to what degree current deep learning architectures for image caption generation capture spatial language. On the basis of the evaluation of examples of generated captions from the literature we argue that systems capture what objects are in the image data but not where these objects are located: the…
The problem of identifying voice commands has always been a challenge due to the presence of noise and variability in speed, pitch, etc. We will compare the efficacies of several neural network architectures for the speech recognition problem. In particular, we will build a model to determine whether a one second audio…