概要
アクティビティ
-
After #ExitAzure, it’s time for 🇮🇳developers to #ExitGoogleMaps! 1 YEAR FREE access for all developers to Ola Maps on Krutrim more than ₹100Cr in…
After #ExitAzure, it’s time for 🇮🇳developers to #ExitGoogleMaps! 1 YEAR FREE access for all developers to Ola Maps on Krutrim more than ₹100Cr in…
Anirudha Vishvakarmaさんが「いいね!」しました
職歴 & 学歴
資格
出版物
-
MILDNet: A Lightweight Single Scaled Deep Ranking Architecture
Arxiv
Multi-scale deep CNN architecture [1, 2, 3] successfully captures both fine and coarse level image descriptors for visual similarity task, but they come up with expensive memory overhead and latency. In this paper, we propose a competing novel CNN architecture, called MILDNet, which merits by being vastly compact (about 3 times). Inspired by the fact that successive CNN layers represent the image with increasing levels of abstraction, we compressed our deep ranking model to a single CNN by…
Multi-scale deep CNN architecture [1, 2, 3] successfully captures both fine and coarse level image descriptors for visual similarity task, but they come up with expensive memory overhead and latency. In this paper, we propose a competing novel CNN architecture, called MILDNet, which merits by being vastly compact (about 3 times). Inspired by the fact that successive CNN layers represent the image with increasing levels of abstraction, we compressed our deep ranking model to a single CNN by coupling activations from multiple intermediate layers along with the last layer. Trained on the famous Street2shop dataset [4], we demonstrate that our approach performs as good as the current state-of-the-art models with only one third of the parameters, model size, training time and significant reduction in inference time. The significance of intermediate layers on image retrieval task has also been shown to be performing on popular datasets Holidays, Oxford, Paris [5]. So even though our experiments are done on ecommerce domain, it is applicable to other domains as well. We further did an ablation study to validate our hypothesis by checking the impact on adding each intermediate layer. With this we also present two more useful variants of MILDNet, a mobile model (12 times smaller) for on-edge devices and a compactly featured model (512-d feature embeddings) for systems with less RAMs and to reduce the ranking cost. Further we present an intuitive way to automatically create a tailored in-house triplet training dataset, which is very hard to create manually. This solution too can also be deployed as an all-inclusive visual similarity solution. Finally, we present our entire production level architecture which currently powers visual similarity at Fynd.
-
Retrieving Similar E-Commerce Images Using Deep Learning
Arxiv
In this paper, we propose a deep convolutional neural network for learning the embeddings of images in order to capture the notion of visual similarity. We present a deep siamese architecture that when trained on positive and negative pairs of images learn an embedding that accurately approximates the ranking of images in order of visual similarity notion. We also implement a novel loss calculation method using an angular loss metrics based on the problems requirement. The final embedding of…
In this paper, we propose a deep convolutional neural network for learning the embeddings of images in order to capture the notion of visual similarity. We present a deep siamese architecture that when trained on positive and negative pairs of images learn an embedding that accurately approximates the ranking of images in order of visual similarity notion. We also implement a novel loss calculation method using an angular loss metrics based on the problems requirement. The final embedding of the image is combined representation of the lower and top-level embeddings. We used fractional distance matrix to calculate the distance between the learned embeddings in n-dimensional space. In the end, we compare our architecture with other existing deep architecture and go on to demonstrate the superiority of our solution in terms of image retrieval by testing the architecture on four datasets. We also show how our suggested network is better than the other traditional deep CNNs used for capturing fine-grained image similarities by learning an optimum embedding.
その他の著者出版物を表示
講習コース
-
Accounting and Financial Management
-
-
Advanced Transport Phenomena
-
-
Business Communications
-
-
Business Ethics
-
-
Chemical Engineering Thermodynamics
-
-
Chemical Process Calculations
-
-
Chemical Processes
-
-
Chemical Reaction Engineering
-
-
Complex Analysis
-
-
Computer Programming and Utilization
-
-
Earth, Environment and Energy
-
-
Engineering Entrepreneurship
-
-
Engineering Mechanics
-
-
Heat Transfer Operations
-
-
International Trade
-
-
Introduction to Electrical Engineering
-
-
Introduction to Numerical Analysis
-
-
Introduction to Sociology
-
-
Mass Transfer Operations
-
-
Material Engineering
-
-
Modern Physics
-
-
Molecular Cell Biology
-
-
Ordinary Differential Equations
-
-
Organizational Behavior and Human Resource Management
-
-
Perspectives in Psychology
-
-
Process Control
-
-
Process Equipment Design
-
-
Process Fluid Mechanics
-
-
Solid Mechanics
-
-
Statistical Mechanics
-
プロジェクト
-
Resonant Deformation of Cancer Cells using Ultrasound driven Oscillating Microbubble
– 現在
Thorough literature review on Microfluidics and it's subcategory Acoustic Microfluidics.
Made a MATLAB code for modeling of the system, verified it with several existing research papers and implemented control strategy to control the deformation of the cell using Simulink.
Next phase is to generalize the model to a more realistic situation and simulation study of the system using COMSOL multiphysics. -
Leader, Covered Scooter Project
– 現在
Led a team of 6 members to design a scooter cover that can filter the polluted air simultaneously, heat/cool it, block rain as well as harmful sun-rays & ensure its comfortability.
Work included modeling of the designs in Autodesk Inventor and testing of their aero-dynamical response using simulation in StarCCM+.その他の作成者 -
Analysis of Polycarbonate manufacture
–
Analyzed and interpreted a flow chart used for industrial manufacturing of polycarbonates. Also identified the nature of the above manufacturing process & did flow process calculations.
受賞歴
-
Dean's List Award
IIT Gandhinagar
Felicitated as a part of Dean’s List for academic excellence in 5th semester.
-
International Robotics Challenge
IIT Bombay
Reached National Finals in International Robotics Challenge Indian Open, Techfest 2012, IIT Bombay.
Built 3 autonomous robots capable of solving a grid and displacing a block.
-
Mean Mechanics Competition
IIT Gandhinagar
Finished 1st in Land, Aqua & Amphibian races of Mean Mechanics competition.
Built a flexible robot from cans capable of traveling in mound tracks and transforming into a boat.
言語
-
English
-
-
Hindi
-
組織
-
AIChE Student Member
-
Anirudhaさんによるその他のアクティビティ
-
India’s first Internet conglomerate may not be Reliance of Tata, but Zomato/Naukri backing 80,000 Cr InfoEdge Reliance is 20L Cr, Tata is 31L Cr…
India’s first Internet conglomerate may not be Reliance of Tata, but Zomato/Naukri backing 80,000 Cr InfoEdge Reliance is 20L Cr, Tata is 31L Cr…
Anirudha Vishvakarmaさんが「いいね!」しました
-
This is what crazy determination, persistence and grit looks like. Many congratulations to team Personate.ai , Akshay Sharma and Rishab Sharma for…
This is what crazy determination, persistence and grit looks like. Many congratulations to team Personate.ai , Akshay Sharma and Rishab Sharma for…
Anirudha Vishvakarmaさんが「いいね!」しました
-
Once you land at a good company with decent pay, do not just keep preparing for interviews and be stuck in this endless cycle ⚡ Engineering is a…
Once you land at a good company with decent pay, do not just keep preparing for interviews and be stuck in this endless cycle ⚡ Engineering is a…
Anirudha Vishvakarmaさんが「いいね!」しました
-
Perplexity is launching business in Japan through partnership with Softbank. It is my great pleasure talking with Mr. Aravind Srinivas and Mr. Dmitry…
Perplexity is launching business in Japan through partnership with Softbank. It is my great pleasure talking with Mr. Aravind Srinivas and Mr. Dmitry…
Anirudha Vishvakarmaさんが「いいね!」しました
-
【Google Japan では現在採用をしています!】 大学時代は HCI…
【Google Japan では現在採用をしています!】 大学時代は HCI…
Anirudha Vishvakarmaさんが「いいね!」しました
-
From Meta, an extensive report on Visual-Language Models(VLM) that connects vision with LLM models. The report explores which model to use, aligning…
From Meta, an extensive report on Visual-Language Models(VLM) that connects vision with LLM models. The report explores which model to use, aligning…
Anirudha Vishvakarmaさんが「いいね!」しました
類似するその他のプロフィール
-
Akash Nagarkar
つながる -
Anuj Arora
つながる -
Rishab Sharma
つながる -
Susumu Yamazaki
ABEJA,inc. - Business Promotion Director / Account Executive GroupManager
つながる -
Pearl Mendonza
Design Director at Rakuten | Design Lead at Fynd | Founder at Love & Glue
つながる -
高橋 佑幸
つながる -
磯崎 陽一
つながる -
杉田 純一
株式会社ABEJA - Project Manager
つながる -
吉田 拓郎
つながる -
Yuhei Okada
つながる
Anirudha Vishvakarmaという名前のその他のユーザー
LinkedInにはAnirudha Vishvakarmaという名前のユーザーが2人います
Anirudha Vishvakarmaという名前のその他のユーザーを表示