G H Raisoni College of Engineering &
Management, Pune
Department Of Data Science
INTERNSHIP REPORT
Internship at
ARDE-DRDO
[company logo]
By:
Yugandhar autade P (63)
Akshay Popale(Roll-4)
Review No:1
Internship Contact Information
Student Name: Yugandhar Autade
Student Email: yugandharautade11@[Link]
Phone No. : 9325557861
Name of Internal Mentor: Mr. Santosh Biradar
Internship Organization ARDE-DRDO
name:
Internship Supervisor: Dr. Harneet Thakur
Internship Supervisor Phone:
INTRODUCTION
For my internship, I had the privilege of working with the Defence Research and Development
Organization (DRDO), the primary R&D wing of India's Ministry of Defence. Established in 1958, its core
mission is to achieve self-reliance in cutting-edge defence technology for our armed forces. DRDO is
responsible for designing and developing a wide array of critical defence systems and technologies. This
experience has provided me with invaluable insight into the vital work being done to strengthen our national
security.
ABOUT COMPANY
The Armament Research & Development Establishment (ARDE), located in Pune, is a premier laboratory
of the Defence Research and Development Organization (DRDO). As the nodal agency for conventional
armaments since 1958, its core mission is to achieve self-reliance in this critical field for India. ARDE is
responsible for the complete development cycle of weapon systems, from research and design to testing and
technology transfer for production. Its work covers a wide spectrum of armament technologies, directly
contributing to the modernization and firepower of the Indian Armed Forces.
WEEK WISE WORK ASSIGNED
Weeks 1 - 2: Comprehensive Onboarding and Literature Review-:
The initial phase was dedicated to a comprehensive project induction and an in-depth study of the
existing literature. My focus was on fully grasping the project's long-term objectives and complex
requirements. This period involved familiarizing myself with the development environment, specific
tools, and the established research methodologies that will be used throughout the internship.
Weeks 3 - 4: Data Acquisition and Exploratory Analysis -:
During this period, I focused on the crucial tasks of data acquisition and curation from various extensive
sources. I implemented robust preprocessing techniques to handle and structure the large-scale dataset.
Subsequently, I conducted a thorough Exploratory Data Analysis (EDA) to identify initial patterns and form
key hypotheses that will guide the project's future development.
Weeks 5 - 6: Methodology Design and Foundational Development -:
Building on the analytical findings, my work shifted towards designing the project's core methodology. I
contributed to outlining the main data processing pipeline and began the initial stages of feature engineering.
This foundational work has helped establish a clear and detailed roadmap for the primary model
development and implementation phases planned for the upcoming months.
WORK ASSIGNED IN DETAIL
The primary objective of my internship project is to design, develop, and evaluate an Artificial Intelligence
model capable of automatically detecting and classifying structural cracks from images. The project leverages
Deep Learning, specifically Convolutional Neural Networks (CNNs), to achieve this. The detailed tasks
assigned to me are as follows.
DETAILED TASKS:
1. Literature Review and Feasibility Study:
Conduct a comprehensive review of existing academic papers and state-of-the-art techniques for CNN-
based structural health monitoring and crack detection to inform the project's direction.
2. Data Acquisition and Curation:
Collect and assemble a large and diverse image dataset of various structural surfaces, containing a
balanced representation of cracked and non-cracked examples.
3. Data Preprocessing and Annotation:
Perform extensive preprocessing on the image dataset, including resizing, normalization, and noise
reduction. Meticulously annotate images by labeling crack locations (e.g., using bounding boxes) and
classifying crack types.
4. Data Augmentation:
Implement data augmentation techniques (such as rotation, flipping, and brightness adjustments) to
artificially expand the dataset, which helps in improving the model's robustness and preventing
overfitting.
5. CNN Model Architecture Selection:
Research, select, and justify a suitable CNN architecture (e.g., VGG, ResNet, YOLO, or a custom
architecture) that is well-suited for the specific tasks of crack detection and classification.
6. Model Development and Implementation:
Implement the chosen CNN model using a deep learning framework like TensorFlow or PyTorch. This
includes coding the network layers, loss functions, and optimization algorithms.
7. Model Training and Hyperparameter Tuning:
Train the CNN model on the prepared dataset. Systematically experiment with and tune
hyperparameters—such as learning rate, batch size, and number of epochs—to achieve optimal
performance.
8. Performance Evaluation and Validation:
Rigorously evaluate the trained model's performance on a separate test dataset using key metrics like
accuracy, precision, recall, F1-score.
9. Result Analysis and Interpretation:
Analyze the model's predictions, interpret its successes and failures, and visualize the results to
understand its decision-making process and identify areas for potential improvement.
[Link] Documentation and Reporting:
Maintain detailed documentation of all project phases, from data collection to model evaluation, and
prepare periodic reports and a final presentation summarizing the methodology, findings, and
conclusions.
CONCLUSION
The initial six weeks of my internship at the Defence Research and Development Organization (DRDO) have
been an immersive and highly educational experience. I have successfully completed the foundational phase
of my project, which has set a clear and promising direction for the work in the upcoming months.
