Summer Research Internship on Vision Tasks

Supervisor

CK Raju

Program Coordinators

Faheema V V

Mulkalla Pavan

Summary of Tasks Completed by Participants and Feedback

• Traffic and Vehicle Detection: The most common tasks involved vehicle detection, often using pre-trained models, for traffic monitoring, surveillance, safety analysis, and management. Specific applications included detecting vehicles in CCTV footage, identifying vehicle flow/density, real-time tracking, detecting ambulances for traffic management, and smart intersection monitoring.
• Road Safety and Infrastructure Monitoring: Several projects focused on improving road safety by detecting hazards and compliance, such as detecting potholes, water logging, open drains, helmet non-compliance, and stray animals.
• Human and Pedestrian Detection/Tracking: Common tasks included detecting humans, pedestrians, and players, as well as more specialized applications like driver drowsiness detection, student attention detection, suspicious activity detection in homes or surveillance videos, and on-device human detection for robot navigation.
• Specialized Detection and Anomaly Analysis: Tasks outside of general traffic/pedestrian monitoring included IV drip flow rate anomaly detection, high-resolution mammographic lesion detection, Parkinson’s Hand Tremor Analysis, electronic component inspection (ECIS), and detecting sickle cells.
• YOLO Model Usage: Many respondents specified using YOLO (e.g., YOLOv8, YOLOv11, YOLO26n) for object detection, dataset preparation, custom image annotation, model fine-tuning, and real-time inference.

• Feedback:
• Positive Learning Experience and Practical Exposure: The internship was highly valued for being a good, informative, structured, and practical hands-on learning experience, significantly strengthening understanding of Computer Vision (CV), Deep Learning, and Machine Learning (ML) concepts. Respondents reported gaining practical exposure to complete CV workflows, including dataset preparation, annotation (using tools like Label Studio), model training (especially YOLO and transfer learning), object detection, semantic segmentation, and real-world deployment challenges.
• Skill Development and Workflow Understanding: The tasks were described as interesting, challenging, well-structured, and straight to the point, helping to improve technical skills in tools such as FFmpeg, Python virtual environments, and GitHub documentation. The experience provided a complete end-to-end understanding of the AI pipeline lifecycle, emphasizing the importance of data quality, preprocessing, and systematic workflow design, rather than just using pre-trained models.
• Suggestions for Improvement: A few respondents suggested that more interactive offline sessions would be engaging and that more detailed beginner guidance and troubleshooting examples could further improve the learning experience. One person requested more offline classes.
• Future Interest and Appreciation: Many expressed satisfaction, curiosity, and a strong desire to pursue further opportunities, research work, or advanced internships in AI/DL/CV with the institute. Gratitude was also expressed for continuous guidance.

Institutes of Participants

• Most Frequent Institutes: Chaitanya Bharathi Institute of Technology (CBIT) and its variations (9 mentions), Keshav Memorial Engineering College and its variations (9 mentions), and C.R. Rao Advanced Institute of Mathematics, Statistics and Computer Science (AIMSCS) and its variations (9 mentions) were the most frequently listed institutions.
• Other Major Contributors: Vallurupalli Nageswara Rao Vignana Jyothi Institute of Engineering and Technology (4 mentions) and Vasavi College of Engineering (4 mentions) also had a notable number of responses.
• Diverse Participation: Responses came from a wide variety of institutions including Anurag University, CMR Technical Campus, Gokaraju Rangaraju Institute of Engineering and Technology, Hyderabad Institute of Technology and Management (HITAM), Indian Institute of Technology Madras, National Institute of Technology Warangal, and University of Hyderabad.
• Specific Engineering Colleges: Several other specific engineering and technology institutes were mentioned, such as Bhoj Reddy Engineering College For Women (3 mentions), Lords Institute of Engineering and Technology (2 mentions), Muffakham Jah College of Engineering & Technology (2 mentions), Neil Gogte Institute of Technology (3 mentions), and Sri Indu Institute of Engineering and Technology (2 mentions).