AI Pioneers: Young Innovators!
Trefo: Anshu Panchamukhi
Problem Statement:
Undernutrition among children under five in India is alarmingly high, with rates of underweight (39-75%), stunting (15.4-74%), and wasting (10.6-42.3%). This malnutrition contributes to approximately 2.3 million child deaths globally each year, representing 41% of deaths in children aged 6-60 months in developing countries. Addressing this issue is critical for improving child health and survival.
Solution:
Tréfo is an AI model leveraging deep neural networks within the computer vision domain to identify malnourished children effectively. By integrating this technology into public surveillance systems and satellites, Tréfo can detect areas with high rates of malnutrition, facilitating timely intervention and resource allocation.
AI Component:
Tréfo employs deep neural networks to analyze images and identify malnourished children. Built using frameworks like TensorFlow and Keras, along with visualization tools such as Matplotlib, this model is implemented in Python. Its automated image analysis capabilities allow for accurate detection of malnutrition, making it adaptable for diverse environments and scalable for both large-scale national assessments and localized evaluations in hospitals and community settings.
SDGs Alignment:
Tréfo directly supports the United Nations' 2nd Sustainable Development Goal (SDG) - "Zero Hunger." By enabling quick identification and response to malnutrition, Tréfo aims to reduce child mortality rates, enhance nutritional health, and promote sustainable food security.
​Microplastic Detection in Human Blood: Himank Arora
Problem Statement: Microplastics, defined as plastic particles smaller than 5 mm, have infiltrated the human food chain and are now present in human blood. These particles pose significant health risks, potentially leading to DNA damage, inflammation, cancer, and other chronic diseases. The urgency for early detection and assessment of microplastic levels in the bloodstream is critical for mitigating these health concerns.
Solution:
To tackle this pressing issue, a deep learning-based software has been developed for the rapid and accurate detection of microplastics in blood samples. Utilizing Red Nile dye to enhance visibility, the software employs a systematic workflow that captures and analyzes images through a CNN. In addition to detection, the software assesses health risks based on the quantity of microplastics identified, thereby providing valuable insights for preventive healthcare.
AI Component:
The core of the solution is a pre-trained CNN model that processes images of stained blood samples. The methodology includes: Image Processing: Implementing color-based segmentation to isolate microplastics.
Thresholding Techniques: Applying RGB thresholding to enhance detection accuracy.
Deep Learning Analysis: The CNN model analyzes the processed images, marking and quantifying the detected microplastics.
The software achieves a detection accuracy of 92%, allowing for effective monitoring of health risks associated with microplastic exposure.
This project aligns with SDG 3: Good Health and Well-Being, as it directly addresses health risks posed by environmental pollutants.
Sporty Coach: Arjun Manocha, Swasti Sharma & Arham Jain
Problem Statement:
Around 60% of parents view sports as a financial strain, and many youths lack access to sports academies or facilities, limiting their ability to pursue their passion. 70% of the 40 million youths who participate in sports annually drop out before age 13, mainly due to financial constraints and lack of resources.
Solution:
Sporty Coach is an affordable, AI-powered digital sports academy that offers personalized training and fitness plans. The app uses AI and AR to provide:
-
Tailored fitness and meal plans for each sport.
-
Daily practice routines and progress tracking.
-
Self-analysis tools and the ability to share gameplay videos.
This solution makes sports training accessible to all, regardless of financial limitations.
AI Components:
-
Pose Detection: Analyzes body posture and provides feedback.
-
Motion Classification: Tracks rest and motion states for optimal training.
-
Personalized Plans: Suggests fitness exercises and meal plans based on sport.
-
Speech Feedback: Converts feedback into speech for real-time guidance.
SDGs Alignment:
-
Quality Education (SDG 4)
-
Decent Work and Economic Growth (SDG 8)
-
Reduced Inequality (SDG 10)
-
Good Health and Well-Being (SDG 3)
Brainy: Dimple Dinkar Patil
Problem Statement:
Alzheimer's disease, accounting for 60-80% of dementia cases, primarily affects individuals aged 65 and older. Early symptoms are often misinterpreted as normal aging, delaying diagnosis and treatment. This leads to worsening cognitive decline and loss of quality of life for patients and their families. In India alone, an estimated 5.3 million people above 60 suffer from dementia, highlighting a significant need for timely intervention and support.
Solution:
We propose an innovative mobile application designed for the early detection of Alzheimer's disease and enhanced care for patients. The app includes:
SAGE Test: A self-administered 15-minute cognitive assessment to identify early signs of impairment. Results are analyzed by experts, who will recommend further testing if necessary.
Games and Quizzes: Engaging activities to stimulate cognitive function and enhance memory, fostering meaningful interactions with caregivers and family.
Location Tracking: A feature for patients to share their live location with family members, ensuring safety and peace of mind.
Emergency Assistance: Quick access to call an ambulance in case of emergencies.
Expert Consultation: A platform for users to ask questions and book appointments with Alzheimer's specialists.
AI Component:
The app leverages AI to analyze user responses in the SAGE Test, providing personalized feedback and recommendations. Machine learning algorithms can also enhance game personalization, adapting difficulty levels based on user performance to maintain engagement and cognitive stimulation.
SDGs Alignment:
SDG 3: Good Health and Well-being
SDG 4: Quality Education
SDG 10: Reduced Inequality
SDG 17: Partnerships for the Goals
FlowGuard: Pradyun Koduru & Abhigna K
Problem Statement:
1. Many times, man hole lids are left open risking the lives of people. The issue is more prominent during rainy season when open manholes are not visible, and individuals may fall inside and die. According to some estimates, at least 2 die each day in India due to open pits and man holes.
2. Additionally, there is no proactive notification, prediction or tracking of water levels of man holes in India, especially when multiple man holes are getting filled quickly in an area, which can be a sign of floods. 3. Finally, humans still enter man holes to clean them (in spite of supreme court banning this practice) due to lack of simple and effective tools. Most of the current machinery is heavy and difficult to use or transport in case of heavy floods or rains.
Solution:
The system addresses these issues through a multi-faceted approach:
a. Safety Alerts: IR sensors on manhole lids detect and alert if lids are open, providing sound or light warnings.
b. Water Level Monitoring: Moisture sensors track water levels in manholes to monitor and display current conditions on a web page.
c. Cleaning Robot: A small robot equipped with tools like rotating blades, centrifugal force mechanisms, an electromagnet, a water motor, and a drill to effectively clean manholes.
d. Predictive Analytics: Machine learning algorithms, using logistic regression, predict potential manhole overflow or flooding based on historical data, weather forecasts, and current conditions.
AI Component:
The system uses machine learning techniques, specifically logistic regression, to predict manhole overflow risks. This model is trained with historical data including rain forecasts, temperature, current water levels, and regional height to provide accurate predictions of potential flooding events.
SDGs Alignment:
Good Health and Well-Being (SDG 3)
Industry, Innovation, and Infrastructure (SDG 9)
FemCare: Prameya Mohanty, Upasana Pradhan & Sushree Aayushi
Problem Statement:
The rise in gynecological diseases, including cervical cancer, breast cancer, ovarian cancer, and polycystic ovarian syndrome (PCOS), poses a significant public health challenge globally, particularly in India. With only about 30,000 gynecologists serving over 600 million women in India, there is a critical shortage of healthcare providers. This results in a ratio of one gynecologist for every 20,000 women, leading to significant delays in diagnosis and treatment. Cervical cancer alone accounts for 604,000 new cases and 342,000 deaths annually worldwide, with India contributing approximately 30% of this burden.
Solution:
FemCare aims to transform gynecological healthcare by leveraging advanced technology to address these challenges. The platform implements:
Advanced Predictive Algorithms: Utilizing machine learning to analyze patient data for early diagnosis and risk assessments of cervical cancer, breast cancer, ovarian cancer, and PCOS.
Efficient Sample Processing: Streamlining the collection and assessment of samples to prioritize high-risk cases for expert review.
Chatbot Integration: Providing a chatbot for mental health support and education on gynecological health issues.
Educational Resources: Offering localized content to enhance awareness among women, particularly in rural areas.
Community Support: Developing forums for shared experiences and professional advice to further enhance emotional and informational support.
AI Component:
FemCare employs sophisticated machine learning algorithms to analyze vast amounts of patient data. These algorithms facilitate accurate risk assessments and early diagnoses, enabling timely medical interventions. The AI system also optimizes resource allocation by prioritizing high-risk cases, thus enhancing overall healthcare efficiency and patient outcomes.
SDGs Alignment:
SDG 3: Good Health and Well-being
VerifiAi- A Deep Learning Approach for Deepfake Detection: Sujan Saitej
Problem Statement
With the rise of deepfake technology, distinguishing between real and manipulated videos has become increasingly challenging. Deepfakes pose significant risks, including misinformation, identity theft, and erosion of trust in media. There is a pressing need for reliable detection tools to combat the potential harm caused by these deceptive videos.
Solution
VerifiAi addresses this problem through a web application that leverages advanced deep learning techniques, specifically ResNeXt for feature extraction and Long Short-Term Memory (LSTM) networks for temporal modeling. Users can upload videos to receive immediate predictions on their authenticity, accompanied by a confidence score that reflects the reliability of the detection.
AI Component
Feature Extraction: The ResNeXt model robustly extracts visual features from individual video frames, identifying crucial elements that differentiate real content from deepfakes.
Temporal Modeling: The LSTM network analyzes the sequence of frames to capture temporal dynamics and detect inconsistencies that may indicate manipulation.
Classification: The combined model outputs predictions—either real or fake—along with a confidence score, providing users with clear insights into the video's authenticity.
SDGs Alignment:
VerifiAi contributes to several SDGs:
SDG 16: Peace, Justice, and Strong Institutions
SDG 9: Industry, Innovation, and Infrastructure
SDG 4: Quality Education
Food Storage: Saanvi Katoch
Problem Statement:
Globally, one-third of food produced is wasted, costing the world economy approximately $750 billion. India contributes significantly to this issue, particularly through pre- and post-harvest waste in cereals and pulses. The preservation of food is crucial, as temperature and humidity directly influence the growth of microorganisms that can reduce shelf life. Farmers, especially in traditional agricultural communities, often rely on outdated techniques for grain storage, leading to substantial losses.
Solution:
This project aims to modernize grain storage practices by utilizing technology to monitor and predict the shelf life food based on environmental conditions. In Palampur, Himachal Pradesh, where agricultural traditions are strong, we have integrated traditional knowledge with modern technology.
AI Components:
Data Collection and Processing: The DHT11 sensor provides real-time environmental data, which is processed by Arduino to display current conditions and transmit them to the Jupyter Notebook.
Machine Learning Prediction: The Decision Tree Regressor analyzes the data to generate predictions about the shelf life of stored grains based on the monitored temperature and humidity.
Notification System: The Arduino integrates with email services to notify users, enhancing the accessibility of vital storage information. SDGs Alignment:
SDG 2: Zero Hunger
SDG 12: Responsible Consumption and Production
SDG 13: Climate Action
SDG 1: No Hunger