Shanmuganathan Raman

Prajwal Singh, Dwip Dalal, Gautam Vashishtha et al.

Decoding the human brain has been a hallmark of neuroscientists and Artificial Intelligence researchers alike. Reconstruction of visual images from brain Electroencephalography (EEG) signals has garnered a lot of interest due to its applications in brain-computer interfacing. This study proposes a two-stage method where the first step is to obtain EEG-derived features for robust learning of deep representations and subsequently utilize the learned representation for image generation and classification. We demonstrate the generalizability of our feature extraction pipeline across three different datasets using deep-learning architectures with supervised and contrastive learning methods. We have performed the zero-shot EEG classification task to support the generalizability claim further. We observed that a subject invariant linearly separable visual representation was learned using EEG data alone in an unimodal setting that gives better k-means accuracy as compared to a joint representation learning between EEG and images. Finally, we propose a novel framework to transform unseen images into the EEG space and reconstruct them with approximation, showcasing the potential for image reconstruction from EEG signals. Our proposed image synthesis method from EEG shows 62.9% and 36.13% inception score improvement on the EEGCVPR40 and the Thoughtviz datasets, which is better than state-of-the-art performance in GAN.

Dwip Dalal, Gautam Vashishtha, Prajwal Singh et al.

Digital imaging aims to replicate realistic scenes, but Low Dynamic Range (LDR) cameras cannot represent the wide dynamic range of real scenes, resulting in under-/overexposed images. This paper presents a deep learning-based approach for recovering intricate details from shadows and highlights while reconstructing High Dynamic Range (HDR) images. We formulate the problem as an image-to-image (I2I) translation task and propose a conditional Denoising Diffusion Probabilistic Model (DDPM) based framework using classifier-free guidance. We incorporate a deep CNN-based autoencoder in our proposed framework to enhance the quality of the latent representation of the input LDR image used for conditioning. Moreover, we introduce a new loss function for LDR-HDR translation tasks, termed Exposure Loss. This loss helps direct gradients in the opposite direction of the saturation, further improving the results' quality. By conducting comprehensive quantitative and qualitative experiments, we have effectively demonstrated the proficiency of our proposed method. The results indicate that a simple conditional diffusion-based method can replace the complex camera pipeline-based architectures.

Ashish Tiwari, Shanmuganathan Raman

Photometric stereo, a problem of recovering 3D surface normals using images of an object captured under different lightings, has been of great interest and importance in computer vision research. Despite the success of existing traditional and deep learning-based methods, it is still challenging due to: (i) the requirement of three or more differently illuminated images, (ii) the inability to model unknown general reflectance, and (iii) the requirement of accurate 3D ground truth surface normals and known lighting information for training. In this work, we attempt to address an under-explored problem of photometric stereo using just two differently illuminated images, referred to as the PS2 problem. It is an intermediate case between a single image-based reconstruction method like Shape from Shading (SfS) and the traditional Photometric Stereo (PS), which requires three or more images. We propose an inverse rendering-based deep learning framework, called DeepPS2, that jointly performs surface normal, albedo, lighting estimation, and image relighting in a completely self-supervised manner with no requirement of ground truth data. We demonstrate how image relighting in conjunction with image reconstruction enhances the lighting estimation in a self-supervised setting.

Aadesh Desai, Saagar Parikh, Seema Kumari et al.

Point cloud segmentation and classification are some of the primary tasks in 3D computer vision with applications ranging from augmented reality to robotics. However, processing point clouds using deep learning-based algorithms is quite challenging due to the irregular point formats. Voxelization or 3D grid-based representation are different ways of applying deep neural networks to this problem. In this paper, we propose PointResNet, a residual block-based approach. Our model directly processes the 3D points, using a deep neural network for the segmentation and classification tasks. The main components of the architecture are: 1) residual blocks and 2) multi-layered perceptron (MLP). We show that it preserves profound features and structural information, which are useful for segmentation and classification tasks. The experimental evaluations demonstrate that the proposed model produces the best results for segmentation and comparable results for classification in comparison to the conventional baselines.

Zeeshan Khan, Parth Shettiwar, Mukul Khanna et al.

Due to hardware constraints, standard off-the-shelf digital cameras suffers from low dynamic range (LDR) and low frame per second (FPS) outputs. Previous works in high dynamic range (HDR) video reconstruction uses sequence of alternating exposure LDR frames as input, and align the neighbouring frames using optical flow based networks. However, these methods often result in motion artifacts in challenging situations. This is because, the alternate exposure frames have to be exposure matched in order to apply alignment using optical flow. Hence, over-saturation and noise in the LDR frames results in inaccurate alignment. To this end, we propose to align the input LDR frames using a pre-trained video frame interpolation network. This results in better alignment of LDR frames, since we circumvent the error-prone exposure matching step, and directly generate intermediate missing frames from the same exposure inputs. Furthermore, it allows us to generate high FPS HDR videos by recursively interpolating the intermediate frames. Through this work, we propose to use video frame interpolation for HDR video reconstruction, and present the first method to generate high FPS HDR videos. Experimental results demonstrate the efficacy of the proposed framework against optical flow based alignment methods, with an absolute improvement of 2.4 PSNR value on standard HDR video datasets [1], [2] and further benchmark our method for high FPS HDR video generation.

Ashish Tiwari, Satoshi Ikehata, Shanmuganathan Raman

Photometric stereo typically demands intricate data acquisition setups involving multiple light sources to recover surface normals accurately. In this paper, we propose MERLiN, an attention-based hourglass network that integrates single image-based inverse rendering and relighting within a single unified framework. We evaluate the performance of photometric stereo methods using these relit images and demonstrate how they can circumvent the underlying challenge of complex data acquisition. Our physically-based model is trained on a large synthetic dataset containing complex shapes with spatially varying BRDF and is designed to handle indirect illumination effects to improve material reconstruction and relighting. Through extensive qualitative and quantitative evaluation, we demonstrate that the proposed framework generalizes well to real-world images, achieving high-quality shape, material estimation, and relighting. We assess these synthetically relit images over photometric stereo benchmark methods for their physical correctness and resulting normal estimation accuracy, paving the way towards single-shot photometric stereo through physically-based relighting. This work allows us to address the single image-based inverse rendering problem holistically, applying well to both synthetic and real data and taking a step towards mitigating the challenge of data acquisition in photometric stereo.

Aalok Gangopadhyay, Abhinav Narayan Harish, Prajwal Singh et al.

We have proposed a self-supervised deep learning framework for solving the mesh blending problem in scenarios where the meshes are not in correspondence. To solve this problem, we have developed Red-Blue MPNN, a novel graph neural network that processes an augmented graph to estimate the correspondence. We have designed a novel conditional refinement scheme to find the exact correspondence when certain conditions are satisfied. We further develop a graph neural network that takes the aligned meshes and the time value as input and fuses this information to process further and generate the desired result. Using motion capture datasets and human mesh designing software, we create a large-scale synthetic dataset consisting of temporal sequences of human meshes in motion. Our results demonstrate that our approach generates realistic deformation of body parts given complex inputs.

Ashish Tiwari, Sresth Tosniwal, Shanmuganathan Raman

Graph convolution networks (GCNs) have been enormously successful in learning representations over several graph-based machine learning tasks. Specific to learning rich node representations, most of the methods have solely relied on the homophily assumption and have shown limited performance on the heterophilous graphs. While several methods have been developed with new architectures to address heterophily, we argue that by learning graph representations across two spaces i.e., topology and feature space GCNs can address heterophily. In this work, we experimentally demonstrate the performance of the proposed GCN framework over semi-supervised node classification task on both homophilous and heterophilous graph benchmarks by learning and combining representations across the topological and the feature spaces.

Sayak Dutta, Harish Katti, Shashikant Verma et al.

Animal tracking and pose estimation systems, such as STEP (Simultaneous Tracking and Pose Estimation) and ViTPose, experience substantial performance drops when processing images and videos with cage structures and systematic occlusions. We present a three-stage preprocessing pipeline that addresses this limitation through: (1) cage segmentation using a Gabor-enhanced ResNet-UNet architecture with tunable orientation filters, (2) cage inpainting using CRFill for content-aware reconstruction of occluded regions, and (3) evaluation of pose estimation and tracking on the uncaged frames. Our Gabor-enhanced segmentation model leverages orientation-aware features with 72 directional kernels to accurately identify and segment cage structures that severely impair the performance of existing methods. Experimental validation demonstrates that removing cage occlusions through our pipeline enables pose estimation and tracking performance comparable to that in environments without occlusions. We also observe significant improvements in keypoint detection accuracy and trajectory consistency.

Anupam Sharma, Harish Katti, Prajwal Singh et al.

An electroencephalogram (EEG) records the spatially averaged electrical activity of neurons in the brain, measured from the human scalp. Prior studies have explored EEG-based classification of objects or concepts, often for passive viewing of briefly presented image or video stimuli, with limited classes. Because EEG exhibits a low signal-to-noise ratio, recognizing fine-grained representations across a large number of classes remains challenging; however, abstract-level object representations may exist. In this work, we investigate whether EEG captures object representations across multiple hierarchical levels, and propose episodic analysis, in which a Machine Learning (ML) model is evaluated across various, yet related, classification tasks (episodes). Unlike prior episodic EEG studies that rely on fixed or randomly sampled classes of equal cardinality, we adopt hierarchy-aware episode sampling using WordNet to generate episodes with variable classes of diverse hierarchy. We also present the largest episodic framework in the EEG domain for detecting observed text from EEG signals in the PEERS dataset, comprising $931538$ EEG samples under $1610$ object labels, acquired from $264$ human participants (subjects) performing controlled cognitive tasks, enabling the study of neural dynamics underlying perception, decision-making, and performance monitoring. We examine how the semantic abstraction level affects classification performance across multiple learning techniques and architectures, providing a comprehensive analysis. The models tend to improve performance when the classification categories are drawn from higher levels of the hierarchy, suggesting sensitivity to abstraction. Our work highlights abstraction depth as an underexplored dimension of EEG decoding and motivates future research in this direction.

Ashish Tiwari, Shanmuganathan Raman

We present a novel inverse rendering-based framework to estimate the 3D shape (per-pixel surface normals and depth) of objects and scenes from single-view polarization images, the problem popularly known as Shape from Polarization (SfP). The existing physics-based and learning-based methods for SfP perform under certain restrictions, i.e., (a) purely diffuse or purely specular reflections, which are seldom in the real surfaces, (b) availability of the ground truth surface normals for direct supervision that are hard to acquire and are limited by the scanner's resolution, and (c) known refractive index. To overcome these restrictions, we start by learning to separate the partially-polarized diffuse and specular reflection components, which we call reflectance cues, based on a modified polarization reflection model and then estimate shape under mixed polarization through an inverse-rendering based self-supervised deep learning framework called SS-SfP, guided by the polarization data and estimated reflectance cues. Furthermore, we also obtain the refractive index as a non-linear least squares solution. Through extensive quantitative and qualitative evaluation, we establish the efficacy of the proposed framework over simple single-object scenes from DeepSfP dataset and complex in-the-wild scenes from SPW dataset in an entirely self-supervised setting. To the best of our knowledge, this is the first learning-based approach to address SfP under mixed polarization in a completely self-supervised framework.

Ashish Tiwari, Satyam Bhardwaj, Yash Bachwana et al.

Estimating scattering parameters of heterogeneous media from images is a severely under-constrained and challenging problem. Most of the existing approaches model BSSRDF either through an analysis-by-synthesis approach, approximating complex path integrals, or using differentiable volume rendering techniques to account for heterogeneity. However, only a few studies have applied learning-based methods to estimate subsurface scattering parameters, but they assume homogeneous media. Interestingly, no specific distribution is known to us that can explicitly model the heterogeneous scattering parameters in the real world. Notably, procedural noise models such as Perlin and Fractal Perlin noise have been effective in representing intricate heterogeneities of natural, organic, and inorganic surfaces. Leveraging this, we first create HeteroSynth, a synthetic dataset comprising photorealistic images of heterogeneous media whose scattering parameters are modeled using Fractal Perlin noise. Furthermore, we propose Tensorial Inverse Scattering (TensoIS), a learning-based feed-forward framework to estimate these Perlin-distributed heterogeneous scattering parameters from sparse multi-view image observations. Instead of directly predicting the 3D scattering parameter volume, TensoIS uses learnable low-rank tensor components to represent the scattering volume. We evaluate TensoIS on unseen heterogeneous variations over shapes from the HeteroSynth test set, smoke and cloud geometries obtained from open-source realistic volumetric simulations, and some real-world samples to establish its effectiveness for inverse scattering. Overall, this study is an attempt to explore Perlin noise distribution, given the lack of any such well-defined distribution in literature, to potentially model real-world heterogeneous scattering in a feed-forward manner.

Ashish Tiwari, Mihir Sutariya, Shanmuganathan Raman

Photometric stereo is a powerful method for obtaining per-pixel surface normals from differently illuminated images of an object. While several methods address photometric stereo with different image (or light) counts ranging from one to two to a hundred, very few focus on learning optimal lighting configuration. Finding an optimal configuration is challenging due to the vast number of possible lighting directions. Moreover, exhaustively sampling all possibilities is impractical due to time and resource constraints. Photometric stereo methods have demonstrated promising performance on existing datasets, which feature limited light directions sparsely sampled from the light space. Therefore, can we optimally utilize these datasets for illumination planning? In this work, we introduce LIPIDS - Learning-based Illumination Planning In Discretized light Space to achieve minimal and optimal lighting configurations for photometric stereo under arbitrary light distribution. We propose a Light Sampling Network (LSNet) that optimizes lighting direction for a fixed number of lights by minimizing the normal loss through a normal regression network. The learned light configurations can directly estimate surface normals during inference, even using an off-the-shelf photometric stereo method. Extensive qualitative and quantitative analyses on synthetic and real-world datasets show that photometric stereo under learned lighting configurations through LIPIDS either surpasses or is nearly comparable to existing illumination planning methods across different photometric stereo backbones.

Shashikant Verma, Shanmuganathan Raman

Despite multiple view consistency offered by 3D-aware GAN techniques, the resulting images often lack the capacity for localized editing. In response, generative radiance manifolds emerge as an efficient approach for constrained point sampling within volumes, effectively reducing computational demands and enabling the learning of fine details. This work introduces SemFaceEdit, a novel method that streamlines the appearance and geometric editing process by generating semantic fields on generative radiance manifolds. Utilizing latent codes, our method effectively disentangles the geometry and appearance associated with different facial semantics within the generated image. In contrast to existing methods that can change the appearance of the entire radiance field, our method enables the precise editing of particular facial semantics while preserving the integrity of other regions. Our network comprises two key modules: the Geometry module, which generates semantic radiance and occupancy fields, and the Appearance module, which is responsible for predicting RGB radiance. We jointly train both modules in adversarial settings to learn semantic-aware geometry and appearance descriptors. The appearance descriptors are then conditioned on their respective semantic latent codes by the Appearance Module, facilitating disentanglement and enhanced control. Our experiments highlight SemFaceEdit's superior performance in semantic field-based editing, particularly in achieving improved radiance field disentanglement.

Aalok Gangopadhyay, Shashikant Verma, Shanmuganathan Raman

We present Deep Mesh Denoising Network (DMD-Net), an end-to-end deep learning framework, for solving the mesh denoising problem. DMD-Net consists of a Graph Convolutional Neural Network in which aggregation is performed in both the primal as well as the dual graph. This is realized in the form of an asymmetric two-stream network, which contains a primal-dual fusion block that enables communication between the primal-stream and the dual-stream. We develop a Feature Guided Transformer (FGT) paradigm, which consists of a feature extractor, a transformer, and a denoiser. The feature extractor estimates the local features, that guide the transformer to compute a transformation, which is applied to the noisy input mesh to obtain a useful intermediate representation. This is further processed by the denoiser to obtain the denoised mesh. Our network is trained on a large scale dataset of 3D objects. We perform exhaustive ablation studies to demonstrate that each component in our network is essential for obtaining the best performance. We show that our method obtains competitive or better results when compared with the state-of-the-art mesh denoising algorithms. We demonstrate that our method is robust to various kinds of noise. We observe that even in the presence of extremely high noise, our method achieves excellent performance.

Prajwal Singh, Gautam Vashishtha, Indra Deep Mastan et al.

The success of deep learning in supervised fine-grained recognition for domain-specific tasks relies heavily on expert annotations. The Open-Set for fine-grained Self-Supervised Learning (SSL) problem aims to enhance performance on downstream tasks by strategically sampling a subset of images (the Core-Set) from a large pool of unlabeled data (the Open-Set). In this paper, we propose a novel method, BloomCoreset, that significantly reduces sampling time from Open-Set while preserving the quality of samples in the coreset. To achieve this, we utilize Bloom filters as an innovative hashing mechanism to store both low- and high-level features of the fine-grained dataset, as captured by Open-CLIP, in a space-efficient manner that enables rapid retrieval of the coreset from the Open-Set. To show the effectiveness of the sampled coreset, we integrate the proposed method into the state-of-the-art fine-grained SSL framework, SimCore [1]. The proposed algorithm drastically outperforms the sampling strategy of the baseline in SimCore [1] with a $98.5\%$ reduction in sampling time with a mere $0.83\%$ average trade-off in accuracy calculated across $11$ downstream datasets.

Prajwal Singh, Ashish Tiwari, Gautam Vashishtha et al.

Neural radiance fields (NeRF) have revolutionized photorealistic rendering of novel views for 3D scenes. Despite their growing popularity and efficiency as 3D resources, NeRFs face scalability challenges due to the need for separate models per scene and the cumulative increase in training time for multiple scenes. The potential for incrementally encoding multiple 3D scenes into a single NeRF model remains largely unexplored. To address this, we introduce Continual-Neural Graphics Primitives (C-NGP), a novel continual learning framework that integrates multiple scenes incrementally into a single neural radiance field. Using a generative replay approach, C-NGP adapts to new scenes without requiring access to old data. We demonstrate that C-NGP can accommodate multiple scenes without increasing the parameter count, producing high-quality novel-view renderings on synthetic and real datasets. Notably, C-NGP models all $8$ scenes from the Real-LLFF dataset together, with only a $2.2\%$ drop in PSNR compared to vanilla NeRF, which models each scene independently. Further, C-NGP allows multiple style edits in the same network.

Shashikant Verma, Shanmuganathan Raman

Achieving realistic hair strand synthesis is essential for creating lifelike digital humans, but producing high-fidelity hair strand geometry remains a significant challenge. Existing methods require a complex setup for data acquisition, involving multi-view images captured in constrained studio environments. Additionally, these methods have longer hair volume estimation and strand synthesis times, which hinder efficiency. We introduce PanoHair, a model that estimates head geometry as signed distance fields using knowledge distillation from a pre-trained generative teacher model for head synthesis. Our approach enables the prediction of semantic segmentation masks and 3D orientations specifically for the hair region of the estimated geometry. Our method is generative and can generate diverse hairstyles with latent space manipulations. For real images, our approach involves an inversion process to infer latent codes and produces visually appealing hair strands, offering a streamlined alternative to complex multi-view data acquisition setups. Given the latent code, PanoHair generates a clean manifold mesh for the hair region in under 5 seconds, along with semantic and orientation maps, marking a significant improvement over existing methods, as demonstrated in our experiments.

Akbar Ali, Mahek Vyas, Soumyaratna Debnath et al.

This paper presents COT-AD, a comprehensive Dataset designed to enhance cotton crop analysis through computer vision. Comprising over 25,000 images captured throughout the cotton growth cycle, with 5,000 annotated images, COT-AD includes aerial imagery for field-scale detection and segmentation and high-resolution DSLR images documenting key diseases. The annotations cover pest and disease recognition, vegetation, and weed analysis, addressing a critical gap in cotton-specific agricultural datasets. COT-AD supports tasks such as classification, segmentation, image restoration, enhancement, deep generative model-based cotton crop synthesis, and early disease management, advancing data-driven crop management

Soumyaratna Debnath, Ashish Tiwari, Kaustubh Sadekar et al.

Recent advancements in learning-based methods have opened new avenues for exploring and interpreting art forms, such as shadow art, origami, and sketch art, through computational models. One notable visual art form is 3D Anamorphic Art in which an ensemble of arbitrarily shaped 3D objects creates a realistic and meaningful expression when observed from a particular viewpoint and loses its coherence over the other viewpoints. In this work, we build on insights from 3D Anamorphic Art to perform 3D object arrangement. We introduce RASP, a differentiable-rendering-based framework to arrange arbitrarily shaped 3D objects within a bounded volume via shadow (or silhouette)-guided optimization with an aim of minimal inter-object spacing and near-maximal occupancy. Furthermore, we propose a novel SDF-based formulation to handle inter-object intersection and container extrusion. We demonstrate that RASP can be extended to part assembly alongside object packing considering 3D objects to be "parts" of another 3D object. Finally, we present artistic illustrations of multi-view anamorphic art, achieving meaningful expressions from multiple viewpoints within a single ensemble.

Shashikant Verma, Harish Katti, Soumyaratna Debnath et al.

We introduce STEP, a novel framework utilizing Transformer-based discriminative model prediction for simultaneous tracking and estimation of pose across diverse animal species and humans. We are inspired by the fact that the human brain exploits spatiotemporal continuity and performs concurrent localization and pose estimation despite the specialization of brain areas for form and motion processing. Traditional discriminative models typically require predefined target states for determining model weights, a challenge we address through Gaussian Map Soft Prediction (GMSP) and Offset Map Regression Adapter (OMRA) Modules. These modules remove the necessity of keypoint target states as input, streamlining the process. Our method starts with a known target state in the initial frame of a given video sequence. It then seamlessly tracks the target and estimates keypoints of anatomical importance as output for subsequent frames. Unlike prevalent top-down pose estimation methods, our approach doesn't rely on per-frame target detections due to its tracking capability. This facilitates a significant advancement in inference efficiency and potential applications. We train and validate our approach on datasets encompassing diverse species. Our experiments demonstrate superior results compared to existing methods, opening doors to various applications, including but not limited to action recognition and behavioral analysis.

Soumyaratna Debnath, Harish Katti, Shashikant Verma et al.

While 2D pose estimation has advanced our ability to interpret body movements in animals and primates, it is limited by the lack of depth information, constraining its application range. 3D pose estimation provides a more comprehensive solution by incorporating spatial depth, yet creating extensive 3D pose datasets for animals is challenging due to their dynamic and unpredictable behaviours in natural settings. To address this, we propose a hybrid approach that utilizes rigged avatars and the pipeline to generate synthetic datasets to acquire the necessary 3D annotations for training. Our method introduces a simple attention-based MLP network for converting 2D poses to 3D, designed to be independent of the input image to ensure scalability for poses in natural environments. Additionally, we identify that existing anatomical keypoint detectors are insufficient for accurate pose retargeting onto arbitrary avatars. To overcome this, we present a lookup table based on a deep pose estimation method using a synthetic collection of diverse actions rigged avatars perform. Our experiments demonstrate the effectiveness and efficiency of this lookup table-based retargeting approach. Overall, we propose a comprehensive framework with systematically synthesized datasets for lifting poses from 2D to 3D and then utilize this to re-target motion from wild settings onto arbitrary avatars.

Jatin Kumar, Indra Deep Mastan, Shanmuganathan Raman

In this paper, we present a Fast Motion Deblurring-Conditional Generative Adversarial Network (FMD-cGAN) that helps in blind motion deblurring of a single image. FMD-cGAN delivers impressive structural similarity and visual appearance after deblurring an image. Like other deep neural network architectures, GANs also suffer from large model size (parameters) and computations. It is not easy to deploy the model on resource constraint devices such as mobile and robotics. With the help of MobileNet based architecture that consists of depthwise separable convolution, we reduce the model size and inference time, without losing the quality of the images. More specifically, we reduce the model size by 3-60x compare to the nearest competitor. The resulting compressed Deblurring cGAN faster than its closest competitors and even qualitative and quantitative results outperform various recently proposed state-of-the-art blind motion deblurring models. We can also use our model for real-time image deblurring tasks. The current experiment on the standard datasets shows the effectiveness of the proposed method.

Mrinal Anand, Nidhin Harilal, Chandan Kumar et al.

High dynamic range (HDR) videos provide a more visually realistic experience than the standard low dynamic range (LDR) videos. Despite having significant progress in HDR imaging, it is still a challenging task to capture high-quality HDR video with a conventional off-the-shelf camera. Existing approaches rely entirely on using dense optical flow between the neighboring LDR sequences to reconstruct an HDR frame. However, they lead to inconsistencies in color and exposure over time when applied to alternating exposures with noisy frames. In this paper, we propose an end-to-end GAN-based framework for HDR video reconstruction from LDR sequences with alternating exposures. We first extract clean LDR frames from noisy LDR video with alternating exposures with a denoising network trained in a self-supervised setting. Using optical flow, we then align the neighboring alternating-exposure frames to a reference frame and then reconstruct high-quality HDR frames in a complete adversarial setting. To further improve the robustness and quality of generated frames, we incorporate temporal stability-based regularization term along with content and style-based losses in the cost function during the training procedure. Experimental results demonstrate that our framework achieves state-of-the-art performance and generates superior quality HDR frames of a video over the existing methods.

Prajwal Singh, Kaustubh Sadekar, Shanmuganathan Raman

Point cloud is one of the widely used techniques for representing and storing 3D geometric data. In the past several methods have been proposed for processing point clouds. Methods such as PointNet and FoldingNet have shown promising results for tasks like 3D shape classification and segmentation. This work proposes a tree-structured autoencoder framework to generate robust embeddings of point clouds by utilizing hierarchical information using graph convolution. We perform multiple experiments to assess the quality of embeddings generated by the proposed encoder architecture and visualize the t-SNE map to highlight its ability to distinguish between different object classes. We further demonstrate the applicability of the proposed framework in applications like: 3D point cloud completion and Single image-based 3D reconstruction.

Prarabdh Raipurkar, Rohil Pal, Shanmuganathan Raman

The prime goal of digital imaging techniques is to reproduce the realistic appearance of a scene. Low Dynamic Range (LDR) cameras are incapable of representing the wide dynamic range of the real-world scene. The captured images turn out to be either too dark (underexposed) or too bright (overexposed). Specifically, saturation in overexposed regions makes the task of reconstructing a High Dynamic Range (HDR) image from single LDR image challenging. In this paper, we propose a deep learning based approach to recover details in the saturated areas while reconstructing the HDR image. We formulate this problem as an image-to-image (I2I) translation task. To this end, we present a novel conditional GAN (cGAN) based framework trained in an end-to-end fashion over the HDR-REAL and HDR-SYNTH datasets. Our framework uses an overexposed mask obtained from a pre-trained segmentation model to facilitate the hallucination task of adding details in the saturated regions. We demonstrate the effectiveness of the proposed method by performing an extensive quantitative and qualitative comparison with several state-of-the-art single-image HDR reconstruction techniques.

Kaustubh Sadekar, Ashish Tiwari, Shanmuganathan Raman

While recent learning based methods have been observed to be superior for several vision-related applications, their potential in generating artistic effects has not been explored much. One such interesting application is Shadow Art - a unique form of sculptural art where 2D shadows cast by a 3D sculpture produce artistic effects. In this work, we revisit shadow art using differentiable rendering based optimization frameworks to obtain the 3D sculpture from a set of shadow (binary) images and their corresponding projection information. Specifically, we discuss shape optimization through voxel as well as mesh-based differentiable renderers. Our choice of using differentiable rendering for generating shadow art sculptures can be attributed to its ability to learn the underlying 3D geometry solely from image data, thus reducing the dependence on 3D ground truth. The qualitative and quantitative results demonstrate the potential of the proposed framework in generating complex 3D sculptures that go beyond those seen in contemporary art pieces using just a set of shadow images as input. Further, we demonstrate the generation of 3D sculptures to cast shadows of faces, animated movie characters, and applicability of the framework to sketch-based 3D reconstruction of underlying shapes.

Abhinav Narayan Harish, Rajendra Nagar, Shanmuganathan Raman

Autonomous assembly of objects is an essential task in robotics and 3D computer vision. It has been studied extensively in robotics as a problem of motion planning, actuator control and obstacle avoidance. However, the task of developing a generalized framework for assembly robust to structural variants remains relatively unexplored. In this work, we tackle this problem using a recurrent graph learning framework considering inter-part relations and the progressive update of the part pose. Our network can learn more plausible predictions of shape structure by accounting for priorly assembled parts. Compared to the current state-of-the-art, our network yields up to 10% improvement in part accuracy and up to 15% improvement in connectivity accuracy on the PartNet dataset. Moreover, our resulting latent space facilitates exciting applications such as shape recovery from the point-cloud components. We conduct extensive experiments to justify our design choices and demonstrate the effectiveness of the proposed framework.

Sudhakar Kumawat, Gagan Kanojia, Shanmuganathan Raman

Deep neural networks have enormous representational power which leads them to overfit on most datasets. Thus, regularizing them is important in order to reduce overfitting and enhance their generalization capabilities. Recently, channel shuffle operation has been introduced for mixing channels in group convolutions in resource efficient networks in order to reduce memory and computations. This paper studies the operation of channel shuffle as a regularization technique in deep convolutional networks. We show that while random shuffling of channels during training drastically reduce their performance, however, randomly shuffling small patches between channels significantly improves their performance. The patches to be shuffled are picked from the same spatial locations in the feature maps such that a patch, when transferred from one channel to another, acts as structured noise for the later channel. We call this method "ShuffleBlock". The proposed ShuffleBlock module is easy to implement and improves the performance of several baseline networks on the task of image classification on CIFAR and ImageNet datasets. It also achieves comparable and in many cases better performance than many other regularization methods. We provide several ablation studies on selecting various hyperparameters of the ShuffleBlock module and propose a new scheduling method that further enhances its performance.

Kaustubh Sadekar, Ashish Tiwari, Prajwal Singh et al.

Handwritten document image binarization is challenging due to high variability in the written content and complex background attributes such as page style, paper quality, stains, shadow gradients, and non-uniform illumination. While the traditional thresholding methods do not effectively generalize on such challenging real-world scenarios, deep learning-based methods have performed relatively well when provided with sufficient training data. However, the existing datasets are limited in size and diversity. This work proposes LS-HDIB - a large-scale handwritten document image binarization dataset containing over a million document images that span numerous real-world scenarios. Additionally, we introduce a novel technique that uses a combination of adaptive thresholding and seamless cloning methods to create the dataset with accurate ground truths. Through an extensive quantitative and qualitative evaluation over eight different deep learning based models, we demonstrate the enhancement in the performance of these models when trained on the LS-HDIB dataset and tested on unseen images.

Aalok Gangopadhyay, Prajwal Singh, Shanmuganathan Raman

Automatic extraction of raw data from 2D line plot images is a problem of great importance having many real-world applications. Several algorithms have been proposed for solving this problem. However, these algorithms involve a significant amount of human intervention. To minimize this intervention, we propose APEX-Net, a deep learning based framework with novel loss functions for solving the plot extraction problem. We introduce APEX-1M, a new large scale dataset which contains both the plot images and the raw data. We demonstrate the performance of APEX-Net on the APEX-1M test set and show that it obtains impressive accuracy. We also show visual results of our network on unseen plot images and demonstrate that it extracts the shape of the plots to a great extent. Finally, we develop a GUI based software for plot extraction that can benefit the community at large. For dataset and more information visit https://sites.google.com/view/apexnetpaper/.

Indra Deep Mastan, Shanmuganathan Raman

One of the major challenges of style transfer is the appropriate image features supervision between the output image and the input (style and content) images. An efficient strategy would be to define an object map between the objects of the style and the content images. However, such a mapping is not well established when there are semantic objects of different types and numbers in the style and the content images. It also leads to content mismatch in the style transfer output, which could reduce the visual quality of the results. We propose an object-based style transfer approach, called DeepObjStyle, for the style supervision in the training data-independent framework. DeepObjStyle preserves the semantics of the objects and achieves better style transfer in the challenging scenario when the style and the content images have a mismatch of image features. We also perform style transfer of images containing a word cloud to demonstrate that DeepObjStyle enables an appropriate image features supervision. We validate the results using quantitative comparisons and user studies.

Indra Deep Mastan, Shanmuganathan Raman

We consider the generic deep image enhancement problem where an input image is transformed into a perceptually better-looking image. Recent methods for image enhancement consider the problem by performing style transfer and image restoration. The methods mostly fall into two categories: training data-based and training data-independent (deep internal learning methods). We perform image enhancement in the deep internal learning framework. Our Deep Internal Learning for Image Enhancement framework enhances content features and style features and uses contextual content loss for preserving image context in the enhanced image. We show results on both hazy and noisy image enhancement. To validate the results, we use structure similarity and perceptual error, which is efficient in measuring the unrealistic deformation present in the images. We show that the proposed framework outperforms the relevant state-of-the-art works for image enhancement.

Indra Deep Mastan, Shanmuganathan Raman

Recently, there is a vast interest in developing image feature learning methods that are independent of the training data, such as deep image prior, InGAN, SinGAN, and DCIL. These methods are unsupervised and are used to perform low-level vision tasks such as image restoration, image editing, and image synthesis. In this work, we proposed a new training data-independent framework, called Deep Contextual Features Learning (DeepCFL), to perform image synthesis and image restoration based on the semantics of the input image. The contextual features are simply the high dimensional vectors representing the semantics of the given image. DeepCFL is a single image GAN framework that learns the distribution of the context vectors from the input image. We show the performance of contextual learning in various challenging scenarios: outpainting, inpainting, and restoration of randomly removed pixels. DeepCFL is applicable when the input source image and the generated target image are not aligned. We illustrate image synthesis using DeepCFL for the task of image resizing.

Harshil Jain, Rohit Patil, Indra Deep Mastan et al.

Blind motion deblurring involves reconstructing a sharp image from an observation that is blurry. It is a problem that is ill-posed and lies in the categories of image restoration problems. The training data-based methods for image deblurring mostly involve training models that take a lot of time. These models are data-hungry i.e., they require a lot of training data to generate satisfactory results. Recently, there are various image feature learning methods developed which relieve us of the need for training data and perform image restoration and image synthesis, e.g., DIP, InGAN, and SinGAN. SinGAN is a generative model that is unconditional and could be learned from a single natural image. This model primarily captures the internal distribution of the patches which are present in the image and is capable of generating samples of varied diversity while preserving the visual content of the image. Images generated from the model are very much like real natural images. In this paper, we focus on blind motion deblurring through SinGAN architecture.

Gagan Kanojia, Shanmuganathan Raman

Consider a set of n images of a scene with dynamic objects captured with a static or a handheld camera. Let the temporal order in which these images are captured be unknown. There can be n! possibilities for the temporal order in which these images could have been captured. In this work, we tackle the problem of temporally sequencing the unordered set of images of a dynamic scene captured with a hand-held camera. We propose a convolutional block which captures the spatial information through 2D convolution kernel and captures the temporal information by utilizing the differences present among the feature maps extracted from the input images. We evaluate the performance of the proposed approach on the dataset extracted from a standard action recognition dataset, UCF101. We show that the proposed approach outperforms the state-of-the-art methods by a significant margin. We show that the network generalizes well by evaluating it on a dataset extracted from the DAVIS dataset, a dataset meant for video object segmentation, when the same network was trained with a dataset extracted from UCF101, a dataset meant for action recognition.

Sudhakar Kumawat, Manisha Verma, Yuta Nakashima et al.

Conventional 3D convolutional neural networks (CNNs) are computationally expensive, memory intensive, prone to overfitting, and most importantly, there is a need to improve their feature learning capabilities. To address these issues, we propose spatio-temporal short term Fourier transform (STFT) blocks, a new class of convolutional blocks that can serve as an alternative to the 3D convolutional layer and its variants in 3D CNNs. An STFT block consists of non-trainable convolution layers that capture spatially and/or temporally local Fourier information using a STFT kernel at multiple low frequency points, followed by a set of trainable linear weights for learning channel correlations. The STFT blocks significantly reduce the space-time complexity in 3D CNNs. In general, they use 3.5 to 4.5 times less parameters and 1.5 to 1.8 times less computational costs when compared to the state-of-the-art methods. Furthermore, their feature learning capabilities are significantly better than the conventional 3D convolutional layer and its variants. Our extensive evaluation on seven action recognition datasets, including Something-something v1 and v2, Jester, Diving-48, Kinetics-400, UCF 101, and HMDB 51, demonstrate that STFT blocks based 3D CNNs achieve on par or even better performance compared to the state-of-the-art methods.

Manisha Verma, Sudhakar Kumawat, Yuta Nakashima et al.

Human pose estimation is a well-known problem in computer vision to locate joint positions. Existing datasets for the learning of poses are observed to be not challenging enough in terms of pose diversity, object occlusion, and viewpoints. This makes the pose annotation process relatively simple and restricts the application of the models that have been trained on them. To handle more variety in human poses, we propose the concept of fine-grained hierarchical pose classification, in which we formulate the pose estimation as a classification task, and propose a dataset, Yoga-82, for large-scale yoga pose recognition with 82 classes. Yoga-82 consists of complex poses where fine annotations may not be possible. To resolve this, we provide hierarchical labels for yoga poses based on the body configuration of the pose. The dataset contains a three-level hierarchy including body positions, variations in body positions, and the actual pose names. We present the classification accuracy of the state-of-the-art convolutional neural network architectures on Yoga-82. We also present several hierarchical variants of DenseNet in order to utilize the hierarchical labels.

Chandra Sekhar Ravuri, Rajesh Sureddi, Sathya Veera Reddy Dendi et al.

The process of rendering high dynamic range (HDR) images to be viewed on conventional displays is called tone mapping. However, tone mapping introduces distortions in the final image which may lead to visual displeasure. To quantify these distortions, we introduce a novel no-reference quality assessment technique for these tone mapped images. This technique is composed of two stages. In the first stage, we employ a convolutional neural network (CNN) to generate quality aware maps (also known as distortion maps) from tone mapped images by training it with the ground truth distortion maps. In the second stage, we model the normalized image and distortion maps using an Asymmetric Generalized Gaussian Distribution (AGGD). The parameters of the AGGD model are then used to estimate the quality score using support vector regression (SVR). We show that the proposed technique delivers competitive performance relative to the state-of-the-art techniques. The novelty of this work is its ability to visualize various distortions as quality maps (distortion maps), especially in the no-reference setting, and to use these maps as features to estimate the quality score of tone mapped images.

Sudhakar Kumawat, Shanmuganathan Raman

In this paper, we propose a new convolutional layer called Depthwise-STFT Separable layer that can serve as an alternative to the standard depthwise separable convolutional layer. The construction of the proposed layer is inspired by the fact that the Fourier coefficients can accurately represent important features such as edges in an image. It utilizes the Fourier coefficients computed (channelwise) in the 2D local neighborhood (e.g., 3x3) of each position of the input map to obtain the feature maps. The Fourier coefficients are computed using 2D Short Term Fourier Transform (STFT) at multiple fixed low frequency points in the 2D local neighborhood at each position. These feature maps at different frequency points are then linearly combined using trainable pointwise (1x1) convolutions. We show that the proposed layer outperforms the standard depthwise separable layer-based models on the CIFAR-10 and CIFAR-100 image classification datasets with reduced space-time complexity.

Zeeshan Khan, Mukul Khanna, Shanmuganathan Raman

High dynamic range (HDR) image generation from a single exposure low dynamic range (LDR) image has been made possible due to the recent advances in Deep Learning. Various feed-forward Convolutional Neural Networks (CNNs) have been proposed for learning LDR to HDR representations. To better utilize the power of CNNs, we exploit the idea of feedback, where the initial low level features are guided by the high level features using a hidden state of a Recurrent Neural Network. Unlike a single forward pass in a conventional feed-forward network, the reconstruction from LDR to HDR in a feedback network is learned over multiple iterations. This enables us to create a coarse-to-fine representation, leading to an improved reconstruction at every iteration. Various advantages over standard feed-forward networks include early reconstruction ability and better reconstruction quality with fewer network parameters. We design a dense feedback block and propose an end-to-end feedback network- FHDR for HDR image generation from a single exposure LDR image. Qualitative and quantitative evaluations show the superiority of our approach over the state-of-the-art methods.

Gagan Kanojia, Shanmuganathan Raman

Consider a set of images of a scene consisting of moving objects captured using a hand-held camera. In this work, we propose an algorithm which takes this set of multi-view images as input, detects the dynamic objects present in the scene, and replaces them with the static regions which are being occluded by them. The proposed algorithm scans the reference image in the row-major order at the pixel level and classifies each pixel as static or dynamic. During the scan, when a pixel is classified as dynamic, the proposed algorithm replaces that pixel value with the corresponding pixel value of the static region which is being occluded by that dynamic region. We show that we achieve artifact-free removal of dynamic objects in multi-view images of several real-world scenes. To the best of our knowledge, we propose the first method which simultaneously detects and removes the dynamic objects present in multi-view images.

Indra Deep Mastan, Shanmuganathan Raman

Recently, there is a vast interest in developing methods which are independent of the training samples such as deep image prior, zero-shot learning, and internal learning. The methods above are based on the common goal of maximizing image features learning from a single image despite inherent technical diversity. In this work, we bridge the gap between the various unsupervised approaches above and propose a general framework for image restoration and image retargeting. We use contextual feature learning and internal learning to improvise the structure similarity between the source and the target images. We perform image resize application in the following setups: classical image resize using super-resolution, a challenging image resize where the low-resolution image contains noise, and content-aware image resize using image retargeting. We also provide comparisons to the relevant state-of-the-art methods.

Shubham Kumar Singh, Krishna P Miyapuram, Shanmuganathan Raman

Person re-identification aims to associate images of the same person over multiple non-overlapping camera views at different times. Depending on the human operator, manual re-identification in large camera networks is highly time consuming and erroneous. Automated person re-identification is required due to the extensive quantity of visual data produced by rapid inflation of large scale distributed multi-camera systems. The state-of-the-art works focus on learning and factorize person appearance features into latent discriminative factors at multiple semantic levels. We propose Deep Parallel Feature Consensus Network (DeepPFCN), a novel network architecture that learns multi-scale person appearance features using convolutional neural networks. This model factorizes the visual appearance of a person into latent discriminative factors at multiple semantic levels. Finally consensus is built. The feature representations learned by DeepPFCN are more robust for the person re-identification task, as we learn discriminative scale-specific features and maximize multi-scale feature fusion selections in multi-scale image inputs. We further exploit average and max pooling in separate scale for person-specific task to discriminate features globally and locally. We demonstrate the re-identification advantages of the proposed DeepPFCN model over the state-of-the-art re-identification methods on three benchmark datasets: Market1501, DukeMTMCreID, and CUHK03. We have achieved mAP results of 75.8%, 64.3%, and 52.6% respectively on these benchmark datasets.

Gagan Kanojia, Sudhakar Kumawat, Shanmuganathan Raman

Traditional 3D convolutions are computationally expensive, memory intensive, and due to large number of parameters, they often tend to overfit. On the other hand, 2D CNNs are less computationally expensive and less memory intensive than 3D CNNs and have shown remarkable results in applications like image classification and object recognition. However, in previous works, it has been observed that they are inferior to 3D CNNs when applied on a spatio-temporal input. In this work, we propose a convolutional block which extracts the spatial information by performing a 2D convolution and extracts the temporal information by exploiting temporal differences, i.e., the change in the spatial information at different time instances, using simple operations of shift, subtract and add without utilizing any trainable parameters. The proposed convolutional block has same number of parameters as of a 2D convolution kernel of size nxn, i.e. n^2, and has n times lesser parameters than an nxnxn 3D convolution kernel. We show that the 3D CNNs perform better when the 3D convolution kernels are replaced by the proposed convolutional blocks. We evaluate the proposed convolutional block on UCF101 and ModelNet datasets.

Indra Deep Mastan, Shanmuganathan Raman

Many real-world solutions for image restoration are learning-free and based on handcrafted image priors such as self-similarity. Recently, deep-learning methods that use training data have achieved state-of-the-art results in various image restoration tasks (e.g., super-resolution and inpainting). Ulyanov et al. bridge the gap between these two families of methods (CVPR 18). They have shown that learning-free methods perform close to the state-of-the-art learning-based methods (approximately 1 PSNR). Their approach benefits from the encoder-decoder network. In this paper, we propose a framework based on the multi-level extensions of the encoder-decoder network, to investigate interesting aspects of the relationship between image restoration and network construction independent of learning. Our framework allows various network structures by modifying the following network components: skip links, cascading of the network input into intermediate layers, a composition of the encoder-decoder subnetworks, and network depth. These handcrafted network structures illustrate how the construction of untrained networks influence the following image restoration tasks: denoising, super-resolution, and inpainting. We also demonstrate image reconstruction using flash and no-flash image pairs. We provide performance comparisons with the state-of-the-art methods for all the restoration tasks above.

Gagan Kanojia, Sudhakar Kumawat, Shanmuganathan Raman

Competitive diving is a well recognized aquatic sport in which a person dives from a platform or a springboard into the water. Based on the acrobatics performed during the dive, diving is classified into a finite set of action classes which are standardized by FINA. In this work, we propose an attention guided LSTM-based neural network architecture for the task of diving classification. The network takes the frames of a diving video as input and determines its class. We evaluate the performance of the proposed model on a recently introduced competitive diving dataset, Diving48. It contains over 18000 video clips which covers 48 classes of diving. The proposed model outperforms the classification accuracy of the state-of-the-art models in both 2D and 3D frameworks by 11.54% and 4.24%, respectively. We show that the network is able to localize the diver in the video frames during the dive without being trained with such a supervision.

Sudhakar Kumawat, Manisha Verma, Shanmuganathan Raman

Recognizing facial expressions is one of the central problems in computer vision. Temporal image sequences have useful spatio-temporal features for recognizing expressions. In this paper, we propose a new 3D Convolution Neural Network (CNN) that can be trained end-to-end for facial expression recognition on temporal image sequences without using facial landmarks. More specifically, a novel 3D convolutional layer that we call Local Binary Volume (LBV) layer is proposed. The LBV layer, when used with our newly proposed LBVCNN network, achieve comparable results compared to state-of-the-art landmark-based or without landmark-based models on image sequences from CK+, Oulu-CASIA, and UNBC McMaster shoulder pain datasets. Furthermore, our LBV layer reduces the number of trainable parameters by a significant amount when compared to a conventional 3D convolutional layer. As a matter of fact, when compared to a 3x3x3 conventional 3D convolutional layer, the LBV layer uses 27 times less trainable parameters.

Sudhakar Kumawat, Shanmuganathan Raman

Traditional 3D Convolutional Neural Networks (CNNs) are computationally expensive, memory intensive, prone to overfit, and most importantly, there is a need to improve their feature learning capabilities. To address these issues, we propose Rectified Local Phase Volume (ReLPV) block, an efficient alternative to the standard 3D convolutional layer. The ReLPV block extracts the phase in a 3D local neighborhood (e.g., 3x3x3) of each position of the input map to obtain the feature maps. The phase is extracted by computing 3D Short Term Fourier Transform (STFT) at multiple fixed low frequency points in the 3D local neighborhood of each position. These feature maps at different frequency points are then linearly combined after passing them through an activation function. The ReLPV block provides significant parameter savings of at least, 3^3 to 13^3 times compared to the standard 3D convolutional layer with the filter sizes 3x3x3 to 13x13x13, respectively. We show that the feature learning capabilities of the ReLPV block are significantly better than the standard 3D convolutional layer. Furthermore, it produces consistently better results across different 3D data representations. We achieve state-of-the-art accuracy on the volumetric ModelNet10 and ModelNet40 datasets while utilizing only 11% parameters of the current state-of-the-art. We also improve the state-of-the-art on the UCF-101 split-1 action recognition dataset by 5.68% (when trained from scratch) while using only 15% of the parameters of the state-of-the-art. The project webpage is available at https://sites.google.com/view/lp-3dcnn/home.

Rajendra Nagar, Shanmuganathan Raman

In computer vision and graphics, various types of symmetries are extensively studied since symmetry present in objects is a fundamental cue for understanding the shape and the structure of objects. In this work, we detect the intrinsic reflective symmetry in triangle meshes where we have to find the intrinsically symmetric point for each point of the shape. We establish correspondences between functions defined on the shapes by extending the functional map framework and then recover the point-to-point correspondences. Previous approaches using the functional map for this task find the functional correspondences matrix by solving a non-linear optimization problem which makes them slow. In this work, we propose a closed form solution for this matrix which makes our approach faster. We find the closed-form solution based on our following results. If the given shape is intrinsically symmetric, then the shortest length geodesic between two intrinsically symmetric points is also intrinsically symmetric. If an eigenfunction of the Laplace-Beltrami operator for the given shape is an even (odd) function, then its restriction on the shortest length geodesic between two intrinsically symmetric points is also an even (odd) function. The sign of a low-frequency eigenfunction is the same on the neighboring points. Our method is invariant to the ordering of the eigenfunctions and has the least time complexity. We achieve the best performance on the SCAPE dataset and comparable performance with the state-of-the-art methods on the TOSCA dataset.