Search Results for author:
Found 108 papers, 51 papers with code
DeiT-LT Distillation Strikes Back for Vision Transformer Training on Long-Tailed Datasets
In DeiT-LT, we introduce an efficient and effective way of distillation from CNN via distillation DIST token by using out-of-distribution images and re-weighting the distillation loss to enhance focus on tail classes.
Ranked #1 on
Image Classification
on iNaturalist
(Overall metric)
Balancing Act: Distribution-Guided Debiasing in Diffusion Models
We train Attribute Distribution Predictor (ADP) - a small mlp that maps the latent features to the distribution of attributes.
Exploring Attribute Variations in Style-based GANs using Diffusion Models
We capitalize on disentangled latent spaces of pretrained GANs and train a Denoising Diffusion Probabilistic Model (DDPM) to learn the latent distribution for diverse edits.
Leveraging Vision-Language Models for Improving Domain Generalization in Image Classification
The proposed approach achieves state-of-the-art results on the standard Domain Generalization benchmarks in a black-box teacher setting as well as a white-box setting where the weights of the VLM are accessible.
Ranked #2 on
Domain Generalization
on DomainNet
Domain-Specificity Inducing Transformers for Source-Free Domain Adaptation
We are the first to utilize vision transformers for domain adaptation in a privacy-oriented source-free setting, and our approach achieves state-of-the-art performance on single-source, multi-source, and multi-target benchmarks
Boosting Adversarial Robustness using Feature Level Stochastic Smoothing
Advances in adversarial defenses have led to a significant improvement in the robustness of Deep Neural Networks.
We never go out of Style: Motion Disentanglement by Subspace Decomposition of Latent Space
Real-world objects perform complex motions that involve multiple independent motion components.
Inspecting the Geographical Representativeness of Images from Text-to-Image Models
Recent progress in generative models has resulted in models that produce both realistic as well as relevant images for most textual inputs.
Certified Adversarial Robustness Within Multiple Perturbation Bounds
We further propose a novel \textit{training noise distribution} along with a \textit{regularized training scheme} to improve the certification within both $\ell_1$ and $\ell_2$ perturbation norms simultaneously.
Continual Domain Adaptation through Pruning-aided Domain-specific Weight Modulation
In this paper, we propose to develop a method to address unsupervised domain adaptation (UDA) in a practical setting of continual learning (CL).
NoisyTwins: Class-Consistent and Diverse Image Generation through StyleGANs
We find that one reason for degradation is the collapse of latents for each class in the $\mathcal{W}$ latent space.
Ranked #1 on
Conditional Image Generation
on ImageNet-LT
Few-Shot Domain Adaptation for Low Light RAW Image Enhancement
Enhancing practical low light raw images is a difficult task due to severe noise and color distortions from short exposure time and limited illumination.
Ranked #1 on
Low-Light Image Enhancement
on Canon RAW Low Light
DART: Diversify-Aggregate-Repeat Training Improves Generalization of Neural Networks
Generalization of neural networks is crucial for deploying them safely in the real world.
RMLVQA: A Margin Loss Approach for Visual Question Answering With Language Biases
The first component considers the frequency of answers within a question type in the training data, which addresses the concern of the class-imbalance causing the language biases.
Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data
Real-world datasets exhibit imbalances of varying types and degrees.
Ranked #1 on
Long-tail Learning
on CIFAR-10-LT (ρ=50)
Subsidiary Prototype Alignment for Universal Domain Adaptation
Universal Domain Adaptation (UniDA) deals with the problem of knowledge transfer between two datasets with domain-shift as well as category-shift.
Efficient and Effective Augmentation Strategy for Adversarial Training
We first explain this contrasting behavior by viewing augmentation during training as a problem of domain generalization, and further propose Diverse Augmentation-based Joint Adversarial Training (DAJAT) to use data augmentations effectively in adversarial training.
Scaling Adversarial Training to Large Perturbation Bounds
The presence of images that flip Oracle predictions and those that do not makes this a challenging setting for adversarial robustness.
Towards Efficient and Effective Self-Supervised Learning of Visual Representations
Self-supervision has emerged as a propitious method for visual representation learning after the recent paradigm shift from handcrafted pretext tasks to instance-similarity based approaches.
Learning an Invertible Output Mapping Can Mitigate Simplicity Bias in Neural Networks
To bridge the gap between these two lines of work, we first hypothesize and verify that while SB may not altogether preclude learning complex features, it amplifies simpler features over complex ones.
Improving GANs for Long-Tailed Data through Group Spectral Regularization
Deep long-tailed learning aims to train useful deep networks on practical, real-world imbalanced distributions, wherein most labels of the tail classes are associated with a few samples.
Ranked #1 on
Image Generation
on LSUN
Hierarchical Semantic Regularization of Latent Spaces in StyleGANs
The quality of the generated images is predicated on two assumptions; (a) The richness of the hierarchical representations learnt by the generator, and, (b) The linearity and smoothness of the style spaces.
Concurrent Subsidiary Supervision for Unsupervised Source-Free Domain Adaptation
The prime challenge in unsupervised domain adaptation (DA) is to mitigate the domain shift between the source and target domains.
Source-Free Domain Adaptation
Unsupervised Domain Adaptation
Everything is There in Latent Space: Attribute Editing and Attribute Style Manipulation by StyleGAN Latent Space Exploration
We propose a novel sampling method to sample latent from the manifold, enabling us to generate a diverse set of attribute styles beyond the styles present in the training set.
Segmentation Guided Deep HDR Deghosting
Our motion segmentation guided HDR fusion approach offers significant advantages over existing HDR deghosting methods.
Balancing Discriminability and Transferability for Source-Free Domain Adaptation
Conventional domain adaptation (DA) techniques aim to improve domain transferability by learning domain-invariant representations; while concurrently preserving the task-discriminability knowledge gathered from the labeled source data.
A Closer Look at Smoothness in Domain Adversarial Training
Based on the analysis, we introduce the Smooth Domain Adversarial Training (SDAT) procedure, which effectively enhances the performance of existing domain adversarial methods for both classification and object detection tasks.
Ranked #5 on
Domain Adaptation
on VisDA2017
Towards Data-Free Model Stealing in a Hard Label Setting
In this work, we show that it is indeed possible to steal Machine Learning models by accessing only top-1 predictions (Hard Label setting) as well, without access to model gradients (Black-Box setting) or even the training dataset (Data-Free setting) within a low query budget.
LEAD: Self-Supervised Landmark Estimation by Aligning Distributions of Feature Similarity
Existing works in self-supervised landmark detection are based on learning dense (pixel-level) feature representations from an image, which are further used to learn landmarks in a semi-supervised manner.
Non-Local Latent Relation Distillation for Self-Adaptive 3D Human Pose Estimation
To this end, we cast 3D pose learning as a self-supervised adaptation problem that aims to transfer the task knowledge from a labeled source domain to a completely unpaired target.
Ranked #5 on
Unsupervised 3D Human Pose Estimation
on Human3.6M
Aligning Silhouette Topology for Self-Adaptive 3D Human Pose Recovery
Articulation-centric 2D/3D pose supervision forms the core training objective in most existing 3D human pose estimation techniques.
Uncertainty-Aware Adaptation for Self-Supervised 3D Human Pose Estimation
The advances in monocular 3D human pose estimation are dominated by supervised techniques that require large-scale 2D/3D pose annotations.
Ranked #8 on
Unsupervised 3D Human Pose Estimation
on Human3.6M
Monocular 3D Human Pose Estimation
Unsupervised 3D Human Pose Estimation
+2
Amplitude Spectrum Transformation for Open Compound Domain Adaptive Semantic Segmentation
However, we find that latent features derived from the Fourier-based amplitude spectrum of deep CNN features hold a more tractable mapping with domain discrimination.
Self-Gated Memory Recurrent Network for Efficient Scalable HDR Deghosting
In the SGM cell, the information flow through a gate is controlled by multiplying the gate's output by a function of itself.
S$^3$VAADA: Submodular Subset Selection for Virtual Adversarial Active Domain Adaptation
Unsupervised domain adaptation (DA) methods have focused on achieving maximal performance through aligning features from source and target domains without using labeled data in the target domain.
Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation
Unsupervised domain adaptation (DA) has gained substantial interest in semantic segmentation.
Ranked #4 on
Domain Generalization
on GTA5-to-Cityscapes
Labeled From Unlabeled: Exploiting Unlabeled Data for Few-Shot Deep HDR Deghosting
To derive data for the next stage of training, we propose a novel method for generating corresponding dynamic inputs from the predicted HDRs of unlabeled data.
Class Balancing GAN with a Classifier in the Loop
However, majority of the developments focus on performance of GANs on balanced datasets.
Deep Implicit Surface Point Prediction Networks
Unsigned distance function (UDF) based approaches have been proposed recently as a promising alternative to represent both open and closed shapes.
Your Classifier can Secretly Suffice Multi-Source Domain Adaptation
Thus, we aim to utilize implicit alignment without additional training objectives to perform adaptation.
Domain Adaptation
Multi-Source Unsupervised Domain Adaptation
S3VAADA: Submodular Subset Selection for Virtual Adversarial Active Domain Adaptation
Unsupervised domain adaptation (DA) methods have focused on achieving maximal performance through aligning features from source and target domains without using labeled data in the target domain.
Guided Adversarial Attack for Evaluating and Enhancing Adversarial Defenses
Further, we propose Guided Adversarial Training (GAT), which achieves state-of-the-art performance amongst single-step defenses by utilizing the proposed relaxation term for both attack generation and training.
Completely Self-Supervised Crowd Counting via Distribution Matching
Dense crowd counting is a challenging task that demands millions of head annotations for training models.
Class-Incremental Domain Adaptation
We introduce a practical Domain Adaptation (DA) paradigm called Class-Incremental Domain Adaptation (CIDA).
Appearance Consensus Driven Self-Supervised Human Mesh Recovery
We present a self-supervised human mesh recovery framework to infer human pose and shape from monocular images in the absence of any paired supervision.
Unsupervised Cross-Modal Alignment for Multi-Person 3D Pose Estimation
Our approach not only generalizes to in-the-wild images, but also yields a superior trade-off between speed and performance, compared to prior top-down approaches.
Fusion of Deep and Non-Deep Methods for Fast Super-Resolution of Satellite Images
In the emerging commercial space industry there is a drastic increase in access to low cost satellite imagery.
Saliency-driven Class Impressions for Feature Visualization of Deep Neural Networks
Existing visualization methods develop high confidence images consisting of both background and foreground features.
Learning to Count in the Crowd from Limited Labeled Data
In this work, we focus on reducing the annotation efforts by learning to count in the crowd from limited number of labeled samples while leveraging a large pool of unlabeled data.
Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimation
However, generalizability of human pose estimation models developed using supervision on large-scale in-studio datasets remains questionable, as these models often perform unsatisfactorily on unseen in-the-wild environments.
From Image Collections to Point Clouds with Self-supervised Shape and Pose Networks
We learn both 3D point cloud reconstruction and pose estimation networks in a self-supervised manner, making use of differentiable point cloud renderer to train with 2D supervision.
3D Object Reconstruction From A Single Image
3D Point Cloud Reconstruction
+2
Adversarial Fooling Beyond "Flipping the Label"
Therefore, the metric to quantify the vulnerability of the models should capture the severity of the flipping as well.
Single-step Adversarial training with Dropout Scheduling
In this work, (i) we show that models trained using single-step adversarial training method learn to prevent the generation of single-step adversaries, and this is due to over-fitting of the model during the initial stages of training, and (ii) to mitigate this effect, we propose a single-step adversarial training method with dropout scheduling.
Universal Source-Free Domain Adaptation
1) In the Procurement stage, we aim to equip the model for future source-free deployment, assuming no prior knowledge of the upcoming category-gap and domain-shift.
Towards Inheritable Models for Open-Set Domain Adaptation
Addressing this, we introduce a practical DA paradigm where a source-trained model is used to facilitate adaptation in the absence of the source dataset in future.
Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesis
Camera captured human pose is an outcome of several sources of variation.
Towards Achieving Adversarial Robustness by Enforcing Feature Consistency Across Bit Planes
In this work, we attempt to address this problem by training networks to form coarse impressions based on the information in higher bit planes, and use the lower bit planes only to refine their prediction.
Regularizers for Single-step Adversarial Training
The proposed regularizers mitigate the effect of gradient masking by harnessing on properties that differentiate a robust model from that of a pseudo robust model.
DeGAN : Data-Enriching GAN for Retrieving Representative Samples from a Trained Classifier
We use the available data, that may be an imbalanced subset of the original training dataset, or a related domain dataset, to retrieve representative samples from a trained classifier, using a novel Data-enriching GAN (DeGAN) framework.
FDA: Feature Disruptive Attack
Though Deep Neural Networks (DNN) show excellent performance across various computer vision tasks, several works show their vulnerability to adversarial samples, i. e., image samples with imperceptible noise engineered to manipulate the network's prediction.
UM-Adapt: Unsupervised Multi-Task Adaptation Using Adversarial Cross-Task Distillation
In this paper, we propose UM-Adapt - a unified framework to effectively perform unsupervised domain adaptation for spatially-structured prediction tasks, simultaneously maintaining a balanced performance across individual tasks in a multi-task setting.
GAN-Tree: An Incrementally Learned Hierarchical Generative Framework for Multi-Modal Data Distributions
Despite the remarkable success of generative adversarial networks, their performance seems less impressive for diverse training sets, requiring learning of discontinuous mapping functions.
Locate, Size and Count: Accurately Resolving People in Dense Crowds via Detection
We introduce a detection framework for dense crowd counting and eliminate the need for the prevalent density regression paradigm.
Ranked #7 on
Crowd Counting
on UCF CC 50
Enhancing Salient Object Segmentation Through Attention
Segmenting salient objects in an image is an important vision task with ubiquitous applications.
Zero-Shot Knowledge Distillation in Deep Networks
Without even using any meta-data, we synthesize the Data Impressions from the complex Teacher model and utilize these as surrogates for the original training data samples to transfer its learning to Student via knowledge distillation.
Dense 3D Point Cloud Reconstruction Using a Deep Pyramid Network
Through extensive quantitative and qualitative evaluation on synthetic and real datasets, we demonstrate that DensePCR outperforms the existing state-of-the-art point cloud reconstruction works, while also providing a light-weight and scalable architecture for predicting high-resolution outputs.
BiHMP-GAN: Bidirectional 3D Human Motion Prediction GAN
The discriminator is trained also to regress this extrinsic factor r, which is used alongside with the intrinsic factor (encoded starting pose sequence) to generate a particular pose sequence.
Unsupervised Feature Learning of Human Actions as Trajectories in Pose Embedding Manifold
In this work we propose a novel temporal pose-sequence modeling framework, which can embed the dynamics of 3D human-skeleton joints to a continuous latent space in an efficient manner.
CAPNet: Continuous Approximation Projection For 3D Point Cloud Reconstruction Using 2D Supervision
We consider the task of single image 3D point cloud reconstruction, and aim to utilize multiple foreground masks as our supervisory data to alleviate the need for large scale 3D datasets.
3D-PSRNet: Part Segmented 3D Point Cloud Reconstruction From a Single Image
We propose a mechanism to reconstruct part annotated 3D point clouds of objects given just a single input image.
Object Pose Estimation from Monocular Image using Multi-View Keypoint Correspondence
In this work, we propose a data-efficient method which utilizes the geometric regularity of intraclass objects for pose estimation.
Gray-box Adversarial Training
Adversarial samples are perturbed inputs crafted to mislead the machine learning systems.
Ask, Acquire, and Attack: Data-free UAP Generation using Class Impressions
Given a model, there exist broadly two approaches to craft UAPs: (i) data-driven: that require data, and (ii) data-free: that do not require data samples.
iSPA-Net: Iterative Semantic Pose Alignment Network
Such image comparison based approach also alleviates the problem of data scarcity and hence enhances scalability of the proposed approach for novel object categories with minimal annotation.
Divide and Grow: Capturing Huge Diversity in Crowd Images with Incrementally Growing CNN
Our model starts from a base CNN density regressor, which is trained in equivalence on all types of crowd images.
Ranked #13 on
Crowd Counting
on WorldExpo’10
Top-Down Feedback for Crowd Counting Convolutional Neural Network
But the current state-of-the-art CNN regressors for crowd counting are feedforward and use only limited spatial context to detect people.
3D-LMNet: Latent Embedding Matching for Accurate and Diverse 3D Point Cloud Reconstruction from a Single Image
3D reconstruction from single view images is an ill-posed problem.
Operator-in-the-Loop Deep Sequential Multi-camera Feature Fusion for Person Re-identification
Therefore, target identifications by operator in a subset of cameras cannot be utilized to improve ranking of the target in remaining set of network cameras.
AdaDepth: Unsupervised Content Congruent Adaptation for Depth Estimation
Supervised deep learning methods have shown promising results for the task of monocular depth estimation; but acquiring ground truth is costly, and prone to noise as well as inaccuracies.
Generalizable Data-free Objective for Crafting Universal Adversarial Perturbations
Further, via exploiting simple priors related to the data distribution, our objective remarkably boosts the fooling ability of the crafted perturbations.
DeepFuse: A Deep Unsupervised Approach for Exposure Fusion with Extreme Exposure Image Pairs
To address the above issues, we have gathered a large dataset of multi-exposure image stacks for training and to circumvent the need for ground truth images, we propose an unsupervised deep learning framework for MEF utilizing a no-reference quality metric as loss function.
NAG: Network for Adversary Generation
Our trained generator network attempts to capture the distribution of adversarial perturbations for a given classifier and readily generates a wide variety of such perturbations.
SketchParse : Towards Rich Descriptions for Poorly Drawn Sketches using Multi-Task Hierarchical Deep Networks
We propose SketchParse, the first deep-network architecture for fully automatic parsing of freehand object sketches.
CNN Fixations: An unraveling approach to visualize the discriminative image regions
We demonstrate through a variety of applications that our approach is able to localize the discriminative image locations across different network architectures, diverse vision tasks and data modalities.
Switching Convolutional Neural Network for Crowd Counting
It is observed that the switch relays an image patch to a particular CNN column based on density of crowd.
Ranked #4 on
Crowd Counting
on Venice
Image Denoising via CNNs: An Adversarial Approach
Is it possible to recover an image from its noisy version using convolutional neural networks?
Confidence estimation in Deep Neural networks via density modelling
State-of-the-art Deep Neural Networks can be easily fooled into providing incorrect high-confidence predictions for images with small amounts of adversarial noise.
Fast Feature Fool: A data independent approach to universal adversarial perturbations
In this paper, for the first time, we propose a novel data independent approach to generate image agnostic perturbations for a range of CNNs trained for object recognition.
Deep image representations using caption generators
We demonstrate that, owing to richer supervision provided during the process of training, the features learned by the captioning system perform better than those of CNNs.
Object category understanding via eye fixations on freehand sketches
In this paper, we analyze the results of a free-viewing gaze fixation study conducted on 3904 freehand sketches distributed across 160 object categories.
'Part'ly first among equals: Semantic part-based benchmarking for state-of-the-art object recognition systems
Our results show that the proposed benchmarking procedure enables additional differentiation among state-of-the-art object classifiers in terms of their ability to handle missing content and insufficient object detail.
Training Sparse Neural Networks
Deep neural networks with lots of parameters are typically used for large-scale computer vision tasks such as image classification.
Generalized Dropout
One set of methods in this family, called Dropout++, is a version of Dropout with trainable parameters.
Compensating for Large In-Plane Rotations in Natural Images
This is inspired by the neuro-scientific concept of mental rotation, which humans use to compare pairs of rotated objects.
CrowdNet: A Deep Convolutional Network for Dense Crowd Counting
Our work proposes a novel deep learning framework for estimating crowd density from static images of highly dense crowds.
Saliency Unified: A Deep Architecture for Simultaneous Eye Fixation Prediction and Salient Object Segmentation
Human eye fixations often correlate with locations of salient objects in the scene.
A Taxonomy of Deep Convolutional Neural Nets for Computer Vision
With this new paradigm, every problem in computer vision is now being re-examined from a deep learning perspective.
Learning Neural Network Architectures using Backpropagation
In this work, we introduce the problem of architecture-learning, i. e; learning the architecture of a neural network along with weights.
DeepFix: A Fully Convolutional Neural Network for predicting Human Eye Fixations
Understanding and predicting the human visual attentional mechanism is an active area of research in the fields of neuroscience and computer vision.
Attribute-Graph: A Graph based approach to Image Ranking
We benchmark the performance of our algorithm on the 'rPascal' and 'rImageNet' datasets, which we have created in order to evaluate the ranking performance on complex queries containing multiple objects.
Data-free parameter pruning for Deep Neural Networks
Our experiments in pruning the densely connected layers show that we can remove upto 85\% of the total parameters in an MNIST-trained network, and about 35\% for AlexNet without significantly affecting performance.
Salient Object Detection via Objectness Measure
Salient object detection has become an important task in many image processing applications.
Crowd Flow Segmentation in Compressed Domain using CRF
This work in compressed domain can be easily extended to pixel domain by substituting motion vectors with motion based features like optical flow.
Expresso : A user-friendly GUI for Designing, Training and Exploring Convolutional Neural Networks
With a view to provide a user-friendly interface for designing, training and developing deep learning frameworks, we have developed Expresso, a GUI tool written in Python.
Object Level Deep Feature Pooling for Compact Image Representation
Convolutional Neural Network (CNN) features have been successfully employed in recent works as an image descriptor for various vision tasks.
Freehand Sketch Recognition Using Deep Features
Therefore, analyzing such sparse sketches can aid our understanding of the neuro-cognitive processes involved in visual representation and recognition.
Category-Epitomes : Discriminatively Minimalist Representations for Object Categories
Freehand line sketches are an interesting and unique form of visual representation.
SeamSeg: Video Object Segmentation using Patch Seams
Label propagation is achieved with high fidelity in the critical boundary regions, utilising the proposed patch seams.
