Renu Rameshan

Aryan Garg, Renu M. Rameshan

Navigating dynamic physical environments without obstructing or damaging human assets is of quintessential importance for social robots. In this work, we solve autonomous drone navigation's sub-problem of predicting out-of-domain human and agent trajectories using a deep generative model. Our method: General-PECNet or G-PECNet observes an improvement of 9.5\% on the Final Displacement Error (FDE) on 2020's benchmark: PECNet through a combination of architectural improvements inspired by periodic activation functions and synthetic trajectory (data) augmentations using Hidden Markov Models (HMMs) and Reinforcement Learning (RL). Additionally, we propose a simple geometry-inspired metric for trajectory non-linearity and outlier detection, helpful for the task. Code available at https://github.com/Aryan-Garg/PECNet-Pedestrian-Trajectory-Prediction.git

Shikha Gupta, Soma Chakraborty, Renu Rameshan

For applications like plant disease detection, usually, a model is trained on publicly available data and tested on field data. This means that the test data distribution is not the same as the training data distribution, which affects the classifier performance adversely. We handle this dataset shift by ensuring that the features are learned from disease spots in the leaf or healthy regions, as applicable. This is achieved using a faster Region-based convolutional neural network (RCNN) as one of the solutions and an attention-based network as the other. The average classification accuracies of these classifiers are approximately 95% while evaluated on the test set corresponding to their training dataset. These classifiers also performed equivalently, with an average score of 84% on a dataset not seen during the training phase.

Anika Shrivastava, Renu Rameshan, Samar Agnihotri

Masked Autoencoders (MAEs) achieve impressive performance in image classification tasks, yet the internal representations they learn remain less understood. This work started as an attempt to understand the strong downstream classification performance of MAE. In this process we discover that representations learned with the pretraining and fine-tuning, are quite robust - demonstrating a good classification performance in the presence of degradations, such as blur and occlusions. Through layer-wise analysis of token embeddings, we show that pretrained MAE progressively constructs its latent space in a class-aware manner across network depth: embeddings from different classes lie in subspaces that become increasingly separable. We further observe that MAE exhibits early and persistent global attention across encoder layers, in contrast to standard Vision Transformers (ViTs). To quantify feature robustness, we introduce two sensitivity indicators: directional alignment between clean and perturbed embeddings, and head-wise retention of active features under degradations. These studies help establish the robust classification performance of MAEs.

Mohana Singh, Renu M. Rameshan

Light field technology has increasingly attracted the attention of the research community with its many possible applications. The lenslet array in commercial plenoptic cameras helps capture both the spatial and angular information of light rays in a single exposure. While the resulting high dimensionality of light field data enables its superior capabilities, it also impedes its extensive adoption. Hence, there is a compelling need for efficient compression of light field images. Existing solutions are commonly composed of several separate modules, some of which may not have been designed for the specific structure and quality of light field data. This increases the complexity of the codec and results in impractical decoding runtimes. We propose a new learning-based, disparity-aided model for compression of 4D light field images capable of parallel decoding. The model is end-to-end trainable, eliminating the need for hand-tuning separate modules and allowing joint learning of rate and distortion. The disparity-aided approach ensures the structural integrity of the reconstructed light fields. Comparisons with the state of the art show encouraging performance in terms of PSNR and MS-SSIM metrics. Also, there is a notable gain in the encoding and decoding runtimes. Source code is available at https://moha23.github.io/LF-DAAE.

Anika Shrivastava, Renu Rameshan, Samar Agnihotri

Understanding the latent spaces learned by deep learning models is crucial in exploring how they represent and generate complex data. Autoencoders (AEs) have played a key role in the area of representation learning, with numerous regularization techniques and training principles developed not only to enhance their ability to learn compact and robust representations, but also to reveal how different architectures influence the structure and smoothness of the lower-dimensional non-linear manifold. We strive to characterize the structure of the latent spaces learned by different autoencoders including convolutional autoencoders (CAEs), denoising autoencoders (DAEs), and variational autoencoders (VAEs) and how they change with the perturbations in the input. By characterizing the matrix manifolds corresponding to the latent spaces, we provide an explanation for the well-known observation that the latent spaces of CAE and DAE form non-smooth manifolds, while that of VAE forms a smooth manifold. We also map the points of the matrix manifold to a Hilbert space using distance preserving transforms and provide an alternate view in terms of the subspaces generated in the Hilbert space as a function of the distortion in the input. The results show that the latent manifolds of CAE and DAE are stratified with each stratum being a smooth product manifold, while the manifold of VAE is a smooth product manifold of two symmetric positive definite matrices and a symmetric positive semi-definite matrix.

Samar Agnihotri, Renu Rameshan, Ritwik Ghosal

Lossless image compression is required in various applications to reduce storage or transmission costs of images, while requiring the reconstructed images to have zero information loss compared to the original. Existing lossless image compression methods either have simple design but poor compression performance, or complex design, better performance, but with no performance guarantees. In our endeavor to develop a lossless image compression method with low complexity and guaranteed performance, we argue that compressibility of a color image is essentially derived from the patterns in its spatial structure, intensity variations, and color variations. Thus, we divide the overall design of a lossless image compression scheme into three parts that exploit corresponding redundancies. We further argue that the binarized version of an image captures its fundamental spatial structure. In this first part of our work, we propose a scheme for lossless compression of binary images. The proposed scheme first learns dictionaries of $16\times16$, $8\times8$, $4\times4$, and $2\times 2$ square pixel patterns from various datasets of binary images. It then uses these dictionaries to encode binary images. These dictionaries have various interesting properties that are further exploited to construct an efficient and scalable scheme. Our preliminary results show that the proposed scheme consistently outperforms existing conventional and learning based lossless compression approaches, and provides, on average, as much as $1.5\times$ better performance than a common general purpose lossless compression scheme (WebP), more than $3\times$ better performance than a state of the art learning based scheme, and better performance than a specialized scheme for binary image compression (JBIG2).

Vidya Sunil, Renu M Rameshan

Eyes serve as our primary sensory organs, responsible for processing up to 80\% of our sensory input. However, common visual aberrations like myopia and hyperopia affect a significant portion of the global population. This paper focuses on simulating a Vision Correction Display (VCD) to enhance the visual experience of individuals with various visual impairments. Utilising Blender, we digitally model the functionality of a VCD in correcting refractive errors such as myopia and hyperopia. With these simulations we can see potential improvements in visual acuity and comfort. These simulations provide valuable insights for the design and development of future VCD technologies, ultimately advancing accessibility and usability for individuals with visual challenges.

Krishan Sharma, Shikha Gupta, Renu Rameshan

In this work, we propose a novel convolutional autoencoder based architecture to generate subspace specific feature representations that are best suited for classification task. The class-specific data is assumed to lie in low dimensional linear subspaces, which could be noisy and not well separated, i.e., subspace distance (principal angle) between two classes is very low. The proposed network uses a novel class-specific self expressiveness (CSSE) layer sandwiched between encoder and decoder networks to generate class-wise subspace representations which are well separated. The CSSE layer along with encoder/ decoder are trained in such a way that data still lies in subspaces in the feature space with minimum principal angle much higher than that of the input space. To demonstrate the effectiveness of the proposed approach, several experiments have been carried out on state-of-the-art machine learning datasets and a significant improvement in classification performance is observed over existing subspace based transformation learning methods.

Amit Satish Unde, Renu M. Rameshan

One of the central tasks of multi-object tracking involves learning a distance metric that is consistent with the semantic similarities of objects. The design of an appropriate loss function that encourages discriminative feature learning is among the most crucial challenges in deep neural network-based metric learning. Despite significant progress, slow convergence and a poor local optimum of the existing contrastive and triplet loss based deep metric learning methods necessitates a better solution. In this paper, we propose cosine-margin-contrastive (CMC) and cosine-margin-triplet (CMT) loss by reformulating both contrastive and triplet loss functions from the perspective of cosine distance. The proposed reformulation as a cosine loss is achieved by feature normalization which distributes the learned features on a hypersphere. We then propose the MOTS R-CNN framework for joint multi-object tracking and segmentation, particularly targeted at improving the tracking performance. Specifically, the tracking problem is addressed through deep metric learning based on the proposed loss functions. We propose a scale-invariant tracking by using a multi-layer feature aggregation scheme to make the model robust against object scale variations and occlusions. The MOTS R-CNN achieves the state-of-the-art tracking performance on the KITTI MOTS dataset. We show that the MOTS R-CNN reduces the identity switching by $62\%$ and $61\%$ on cars and pedestrians, respectively in comparison to Track R-CNN.