Deepak NR

Computer Vision, Deep Learning
Data Science and Python 

Medium
GitHub
LinkedIn
Email
About

About Me

Hello, I'm Deepak NR, skilled in Computer Vision, Deep Learning, and Data Science in Python. I'm passionate about AI and hold a track record of streamlining projects, such as Automatic License Plate Recognition and Face Recognition. With hands-on experience in Numpy/Pandas, OpenCV, and more, I deliver tangible results. I'm dedicated to leveraging my skills in Object Detection, Object Tracking, and Optical Character Recognition to drive innovation. Currently, I'm eager to collaborate on impactful AI projects. Let's explore possibilities together!

My Certifications

Advanced Computer Vision with TensorFlow

Explored image tasks: classification, segmentation, object detection. Used transfer learning for detection. Employed models like RCNN, ResNet-50, U-Net, and Mask-RCNN for rubber ducks, numbers, pets, etc. Analyzed predictions with activation and saliency maps, improving AlexNet.

Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization

 

Explored deep learning processes, optimizing neural networks systematically. Learned best practices in training, test sets, bias/variance analysis, regularization, optimization algorithms (e.g., Momentum, Adam), and TensorFlow implementation

Neural Networks and Deep Learning

 

Studied foundational neural networks and deep learning concepts. Explored technological trends, built/train deep neural networks, implemented efficient architectures, and applied deep learning to own projects


Python (Basics)

 

It encompasses subjects such as Scalar Types, Operators, Control Flow, Strings, Collections, Iteration, Modularity, Objects, Types, and Classes

My Work Experience

ElectrifAi Technology Solutions Pvt Ltd

Associate Data Scientist
January 2023-February 2023

• Designed a self-annotation system for annotating object detection data (Raw data)

• Provided technical support to clients and collaborated with cross-functional teams to resolve technical issues

• Actively participate and contribute to the internal data science project initiatives

Capulus Technologies Pvt Ltd

Associate Software Engineer
January 2021-September 2022

• Improved model inference speed by 20% in projects: Automatic License Plate Recognition and Face Recognition, utilizing Nvidia CUDA and CUDNN.

• Designed and custom-trained Object detection (YOLOv4)/OCR frameworks, including CRNN and Tesseract OCR, optimizing deep-learning solutions.

• Assembled 100k+ data points for diverse deep-learning model training, ensuring robust datasets.

• Conducted precise data annotation using Computer Vision Annotation Tool (CVAT) for custom object detection.

• Proficient in various IDEs like Google Colab, Jupiter Lab, PyCharm, and VS Code for seamless development.

• Collaborated with cross-functional teams, addressing 100+ critical technical issues and enhancing project efficiency.

Blog

My Blogs

Case Study: Strategies for Accurate Small Object Detection in Images

One of the most challenging and significant issues in computer vision is the detection of small objects. To reliably identify small things in a video feed or image, you must solve the computer vision problem of small object detection. 


The size of the thing itself is not a requirement. In aerial computer vision, for instance, it’s essential to be able to reliably identify objects even if each one will be small in relation to the scale of the photo.

U-Net Demystified: A Gentle Introduction

The U-Net architecture has redefined the way we approach image segmentation tasks, making it a vital topic for student researchers and beginners in the field. In this comprehensive guide, we will embark on an exciting journey to unravel the intricacies of the U-Net architecture and explore its core components and real-world applications. Whether you’re an aspiring data scientist or a curious novice, this article is designed to equip you with a profound understanding of U-Net’s capabilities and its role in reshaping the field of computer vision.

Potholes and Plain road images classification using CNN(with code)

Roads are essential to transportation, providing a means for people to travel from one place to another. However, maintenance of roads can be challenging, especially with the occurrence of potholes. Potholes are common problems on roads that can cause significant damage to vehicles and even lead to accidents.

Fire and Smoke detection using PyTorch (with full code)

In recent years, artificial intelligence and machine learning advancements have revolutionized various industries, including public safety. One area where these technologies have made significant strides is in fire and smoke detection, which is crucial for early warning systems and efficient emergency response. Fire and smoke detection systems play an important role in preventing disasters and minimizing property damage.

A simple example of using Fast API and Googletrans in Python

FastAPI is a modern, fast web framework for building APIs with Python. It was first released in 2019 and quickly became popular among python developers, who praised its performance and easy-to-use design. FastAPI is built on top of Starlette, a web framework that provides performance comparable to that of other popular frameworks such as Flask and Django.

Embracing Tomorrow’s Deep Learning: Anticipated Trends and Innovations

Deep learning, a subfield of artificial intelligence (AI), has witnessed remarkable advancements in recent years. From image recognition to natural language processing, deep learning models have demonstrated superhuman performance in various tasks. 


Deep learning is poised for even more groundbreaking developments as we venture into the future. In this article, we’ll explore the exciting trends and innovations that will shape the future of deep learning. Plus, we’ll dive into Python code examples to help you get hands-on experience with these cutting-edge technologies.

Contact 

Contact Me