Prøveforelesning og disputas – Ravindra R. Patil / Trial lecture and defense – Ravindra R. Patil

Ravindra R. Patil disputerer for ph.d.-graden i ingeniørvitenskap og vil offentlig forsvare avhandlingen / Ravindra R. Patil will defend his thesis for the PhD degree in Engineering Science:

“Enhancing AI Systems through Representative Dataset, Transfer Learning, and Embedded Vision”.

Avhandlingen er tilgjengelig her / The doctoral thesis is available here.

Auditoriet er åpent for publikum. Disputasen vil også bli strømmet. Opptak av disputasen vil være tilgjengelig i en måned.
The auditorium is open to the public. The defense will be streamed. A recording of the defense will be available for one month.

Prøveforelesningen starter kl. 10:15 / The trial lecture starts at 10:15. Tittel / title:

“Artificial Intelligence for Computer Vision Tasks”.

Disputasen starter kl. 12:15 / The defense starts at 12:15.

Prøveforelesning strømmes her, disputas strømmes her. / The trial lecture will be streamed here, and defense will be streamed here.

Sammendrag av avhandlingen / Summary of the thesis:

Artificial intelligence (AI) is revolutionizing the way we solve complex problems, and in my PhD thesis, I've focused on harnessing the power of AI to address a critical issue: sewer blockages in urban wastewater systems. This work has far-reaching implications for improving wastewater management in cities and beyond.

Why is this important? Sewer blockages are a common and costly problem in urban areas. They can lead to sewage backups, environmental pollution, and significant maintenance expenses. Traditional methods for detecting and addressing these blockages are often slow and reactive. In contrast, AI offers a proactive and efficient solution.

Creating the Right Dataset: S-BIRD. One of the key challenges in AI is having high-quality data to train models effectively. In my research, I tackled this problem head-on by creating the Sewer-Blockages Imagery Recognition Dataset (S-BIRD). This dataset captures real-world sewer blockage scenarios with unprecedented authenticity. It became the foundation for developing accurate and robust AI models.

Training AI Models. I used deep neural networks, a cutting-edge AI technique, to develop detection models for sewer blockages. But to make these models work effectively, they needed to learn from diverse and representative data. S-BIRD played a crucial role in providing this data. The models were trained using transfer learning and fine-tuning techniques, allowing them to adapt to different situations with minimal additional training.

Impressive Results. The results of our experiments were remarkable. Our AI models achieved a high accuracy rate of 96.30% in detecting various types of sewer blockages. This demonstrated not only the effectiveness of the S-BIRD dataset but also the applicability of the techniques used for developing the models.

Real-World Application. But AI is not just about algorithms; it's about solving real-world problems. We took our AI detector, trained on the S-BIRD dataset, and embedded it in a vision-based automation system. This system can now detect sewer blockages in real-time, making urban wastewater management more efficient and reliable.

Looking Ahead. There are exciting avenues for further exploration:

• We can explore additional AI techniques like semantic, instance or panoptic segmentation to enhance detection tasks further.
• Expanding the dataset with more classes and challenging scenarios can make our models even more effective.
• Collaboration with industry partners can help implement these AI-driven solutions on a larger scale, benefiting cities worldwide.

In conclusion, my PhD thesis demonstrates how AI can revolutionize wastewater management by addressing sewer blockages efficiently. With the S-BIRD dataset, advanced AI models, and real-time detection systems, we are paving the way for a cleaner and more sustainable urban environment. This work is not only significant for the field of AI but also for improving the quality of life in our cities.

Veiledere / Supervisors:

Hovedveileder / Main supervisor:

Professor Mohamad Y. Mustafa, UiT The Arctic University of Norway, Department of Building, Energy and Material Technology (IBEM),

Biveiledere / Co-supervisor:

Professor Rajnish K. Calay, UiT The Arctic University of Norway, IBEM,

Dr. Saniya M. Ansari, D Y Patil School of Engineering, Pune, India.

Bedømmelseskomité / Evaluation committee:

• Professor emeritus Elijah Liflyand, Bar Ilan University, Ramat Gan, Israel, 1. opponent,
• Dr. Lily Meng, Engineering Department, School of Physics, Engineering and Computer Science, University of Hertfordshire, England, 2. opponent,
• Professor Natasha Gabatsuyevna Samko, UiT The Arctic University of Norway, internt medlem og komiteens administrator / leader of the committee.

Prøveforelesning og disputas ledes av prodekan for forskning, Svein-Erik Sveen /

The trial lecture and defense are led by Vice Dean of research, Svein-Erik Sveen.

De som ønsker å opponere ex auditorio kan sende e-post til Svein-Erik Sveen / Opponents ex auditorio should contact Svein-Erik Sveen.