Kankanala Sai Varsha

I aspire to work under a recognized & reputed organization for the advancement of my career & personal growth like exposure, knowledge, expertise. Detail-oriented, organized and meticulous employee. Works at fast pace to meet tight deadlines. Enthusiastic team player ready to contribute to company success.

• IBM - Artificial Intelligence & Machine Learning with Python
• Project Development Using Java - WIPRO
• Cultural coordinator for all department programs in our college.
• Rendered the social service called NSS(National Service Scheme) as a volunteer
• Actively Volunteered for WORLD SPACE WEEK celebrations organized by ISRO

• Participated in Kaggle,TCS,shiksha competitions, including the 24hr Hackathon competition, where I developed and trained machine learning models to classify crop health using images.
• Active member of online IT communities, including Kaggle, GitHub, and Reddit, where I engage with other professionals and enthusiasts to learn and share knowledge on SQL/Python/Java applications.
• Attended meetups and conferences focused on Innovative technology, where I networked with professionals and learned about the latest developments in Python/Java/PowerBI applications .
• Mentored students and enthusiasts in AI/ML and Python, guiding them in developing projects related to crop health assessment and agricultural technology.

Major project: HUMAN ACTION RECOGNITION

Human Action Recognition is a computer vision task that involves identifying and classifying human actions in video data. It has numerous applications in fields such as surveillance, human-computer interaction, and healthcare. It can be used to detect anomalous behavior, monitor elderly people, and improve accessibility for people with disabilities.

Techniques used in Human Action Recognition: There are various techniques used in Human Action Recognition, including:

• Traditional machine learning algorithms such as Support Vector Machines (SVMs) and Random Forests.
• Deep learning models such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs).
• Motion history images and optical flow analysis.
• 3D convolutional neural networks (3D CNNs) and spatiotemporal features.

Datasets used in Human Action Recognition: There are several datasets available for Human Action Recognition, including:

• UCF101: A dataset containing 13,320 videos and 101 action categories.
• HMDB51: A dataset containing 7,000 videos and 51 action categories.
• KTH: A dataset containing 600 videos and 6 action categories.
• UCF50: A dataset containing 7,500 videos and 50 action categories.

Developed a Human Action Recognition system using a 3D CNN and spatiotemporal features, achieving an accuracy of 89% on the UCF101 dataset.Implemented a deep learning model for action recognition using TensorFlow, achieving state-of-the-art results on the HMDB51 dataset.

Worked on a research project related to Human Action Recognition, investigating the use of motion history images and optical flow analysis for action recognition.

Intern project: CROP HEALTH ASSESSMENT

Developed a machine learning model to classify crop health using images of crops, achieving an accuracy of 92%.

• Utilized Python and various libraries to preprocess images, extract features, and train the model.
• Technologies used: Python, TensorFlow/Keras, OpenCV, Scikit-learn, Pandas/Numpy.
• Dataset: Utilized the Crop Health dataset from Kaggle, consisting of 10,000 images of crops.
• Skills demonstrated: Machine Learning, Image Processing, Data Analysis, Python Programming.
• Achievements: Achieved an accuracy of 92% in classifying crop health using the developed model, and improved the efficiency of the model by 30% through optimization techniques.