Utkarsh Tomar Email & Phone Number

Rice Disease Classifier: Empowering Filipino Farmers with AIProtecting Filipino Food Security with Smart TechnologyMy Rice Disease Classifier project empowers Filipino farmers with a web app that uses computer vision to diagnose rice diseases quickly and accurately. This innovative tool tackles a critical challenge facing Filipino agriculture, promoting increased rice production and safeguarding food security.Key Highlights:Expert Knowledge: I researched common rice diseases in the Philippines, understanding their symptoms, causes, and management strategies.Building a Robust Dataset: I aggregated and cleaned data from diverse sources, including online databases and image repositories, ensuring a strong foundation for model training.AI-Powered Diagnosis: I developed and trained a CNN model using TensorFlow and Keras, enabling precise rice disease identification based on images.Accessible Solution: I deployed the model on a user-friendly web app built with Flask and Streamlit. Farmers can upload images, receive diagnoses, and access tailored recommendations for treatment and prevention.Impact and Achievements:Enhanced Food Security: Early disease detection helps farmers implement timely interventions, leading to higher rice yields and improved food security.Empowered Farmers: The web app provides readily available knowledge and tools, promoting sustainable agricultural practices and protecting Filipino food culture.Technology for Good: I showcased the positive impact of computer vision and machine learning in solving agricultural challenges faced by Filipino farmers.Accessible for All: The user-friendly interface makes the tool accessible to farmers regardless of their technical expertise, fostering knowledge dissemination and effective disease management.My Skills and Expertise:Data collection, cleaning, and analysisComputer vision and CNN model developmentWeb app development with Flask and StreamlitTensorFlow and Keras proficiency

Zowasel: Empowering Nigerian Farmers with AI-powered Credit and Yield InsightsBridging the Gap in Agricultural Financing and SustainabilityMy work at Zowasel empowered local farmers through a sophisticated Flask API, using machine learning for credit scoring and crop yield prediction. This tackled two key challenges: financing access and resource management.Key Highlights:Data-driven Insights: I analyzed and cleansed data, employing feature engineering for optimal model performance.Model Powerhouse: I built and compared machine learning models for credit scoring and integrated a deep learning model for crop yield prediction, providing a holistic picture for lenders.Seamless Integration: I deployed the API within Zowasel's app, generating reports with predictions and insights for farmers and lenders.Leadership and Collaboration: I led a data collection team of 20+, ensuring high-quality data and communicating outcomes to stakeholders.Impact:Enhanced Financial Inclusion: Improved access to credit for farmers, promoting sustainable practices and economic growth.Data-driven Decision Making: Accurate predictions for informed financial decisions by farmers and lenders.Cross-disciplinary Expertise: Bridged data science and computer vision for a comprehensive agricultural solution.Effective Leadership: Successfully managed a large team and ensured project delivery.

SurplusMap: Automating Power Grid Infrastructure Detection with AIStreamlining Infrastructure Mapping for Green Energy InitiativesMy work at SurplusMap centered around developing a revolutionary Streamlit application that automatically identifies and maps power grid infrastructure (pylons, substations, power stations) from satellite images using computer vision. This innovative solution revolutionizes the manual identification process, saving time, resources, and driving green energy initiatives forward.Key Highlights:Automated Efficiency: I implemented and fine-tuned YOLOv5/YOLT models on a custom-labeled dataset, achieving high accuracy in identifying power grid elements, significantly boosting efficiency.Data Quality Champion: I led a 20+ member team in data pre-processing, ensuring high-quality data for model training and reducing pre-processing time by 50% through effective collaboration.Scalability for Impact: I demonstrated the project's potential to scale to other countries with similar geographical features, supporting green transition efforts through efficient power grid mapping.My Technical Arsenal:Computer Vision: Object detection, image segmentation, image classificationMachine Learning: YOLOv5/YOLT, pre-trained models, fine-tuningData Processing: Data collection, annotation, pre-processingStreamlit App DevelopmentTeam Leadership and CollaborationQuantifiable Achievements:Increased efficiency in power grid infrastructure identification by 20%.Improved object detection accuracy by 40% compared to baseline models.Reduced data pre-processing time by 50% through effective team leadership.My Contribution:Beyond technical expertise, I played a pivotal role in leading the data pre-processing team, ensuring high-quality data that fueled the project's success. My dedication to green initiatives drove me to explore the project's scalability potential, paving the way for wider adoption and impact.

Unveiling Cricket's Nuances: My Video Analysis Internship at MASAs a video data analyst at Mad About Sports (MAS), I dove deep into the fascinating world of bowling and batting techniques, enhancing their analytics offerings. My expertise shone through in:Data Acquisition and Processing: Efficiently navigating diverse video sources, I organized data on various bowling and batting styles.Taxonomy Development: My crowning achievement: a comprehensive taxonomy of all existing bowling variations, with detailed descriptions and player mastery mappings. This valuable resource unlocks nuanced understanding.Elite Player Analysis: I went behind the scenes with elite pacers and spinners, dissecting their attributes, variations, and matchups through meticulous video analysis.Batsman Classification: Recognizing the key role of batting styles, I established a robust classification system based on handedness, back lift, and other key factors, unveiling insights into individual performances against specific bowlers.Clear Communication: My findings translated into comprehensive documents, presentations, and excel sheets, effectively informing both internal and external stakeholders.Impact and Achievements:Elevated Cricket Analytics: My work enriched MAS's offerings with unparalleled insights into diverse bowling and batting techniques.Data Analysis Virtuosity: I showcased mastery in data collection, processing, analysis, and visualization, extracting crucial information from video footage.Taxonomy Innovation: The bowling variations taxonomy stands as a significant contribution to cricket, enhancing understanding and classification.Player Insights Revealed: My analyses shed light on the strategies and techniques of elite players, offering valuable knowledge for coaches and players alike.

Data Science

Bachelor Of Technology - Btech, Information Technology