no code implementations • 30 Aug 2023 • Ujjal Kr Dutta, Aldo Lipani, Chuan Wang, Yukun Hu
Foundation Industries (FIs) constitute glass, metals, cement, ceramics, bulk chemicals, paper, steel, etc.
no code implementations • 20 Aug 2022 • Ujjal Kr Dutta
While deep Embedding Learning approaches have witnessed widespread success in multiple computer vision tasks, the state-of-the-art methods for representing natural images need not necessarily perform well on images from other domains, such as paintings, cartoons, and sketch.
no code implementations • 21 Mar 2022 • Purbarag Pathak Choudhury, Ujjal Kr Dutta, Dhruba Kr Bhattacharyya
A satellite image is a remotely sensed image data, where each pixel represents a specific location on earth.
no code implementations • 6 Dec 2021 • Ujjal Kr Dutta, Sandeep Repakula, Maulik Parmar, Abhinav Ravi
Our framework could be trained with supervisory signals in the form of triplets, that are obtained manually.
no code implementations • 6 Dec 2021 • Ujjal Kr Dutta
We propose a novel, contrastive learning method to align the latent representations of a pair of real and synthetic images, to make the detector robust to the different domains.
no code implementations • 10 May 2021 • Ujjal Kr Dutta, Mehrtash Harandi, Chellu Chandra Sekhar
In this paper, we address the problem of Semi-Supervised DML (SSDML) that tries to learn a metric using a few labeled examples, and abundantly available unlabeled examples.
no code implementations • 17 Apr 2021 • Ujjal Kr Dutta, Sandeep Repakula, Maulik Parmar, Abhinav Ravi
Interestingly, we observed that color variants are essentially manifestations of color jitter based augmentations.
1 code implementation • 26 Aug 2020 • Abhinav Ravi, Sandeep Repakula, Ujjal Kr Dutta, Maulik Parmar
The novelty and strength of our method lies in its capability to recommend similar articles for all the fashion items worn by the model, in addition to the primary article corresponding to the query.
no code implementations • 26 Aug 2020 • Rajdeep Hazra Banerjee, Abhinav Ravi, Ujjal Kr Dutta
The trained model is fine-tuned to generate style-based captions for the target dataset.
no code implementations • 22 Aug 2020 • Ujjal Kr Dutta, Mehrtash Harandi, Chellu Chandra Sekhar
As obtaining class labels in all applications is not feasible, we propose an unsupervised approach that learns a metric without making use of class labels.
no code implementations • 27 Feb 2020 • Ujjal Kr Dutta, Mehrtash Harandi, Chellu Chandra Sekhar
In this paper, we revamp the forgotten classical Semi-Supervised Distance Metric Learning (SSDML) problem from a Riemannian geometric lens, to leverage stochastic optimization within a end-to-end deep framework.
no code implementations • 17 Dec 2019 • Ujjal Kr Dutta, Mehrtash Harandi, Chandra Sekhar Chellu
This restricts their applicability for large datasets in new applications where obtaining labels require extensive manual efforts and domain knowledge.