Titulo Estágio
Continual Learning for Intelligent Biodiversity Detection
Áreas de especialidade
Sistemas Inteligentes
Local do Estágio
CISUC,DEI
Enquadramento
Agriculture is fundamental to the global economy, serving as the primary source of food worldwide. As the global population continues to grow, so does the demand for food, making the efficiency of production increasingly critical.
Animal biodiversity significantly influences crop yields and sustainability. Some animals, such as pests, can damage crops, while others benefit agriculture by controlling these pests, pollinating plants, or enhancing soil quality. Therefore, monitoring biodiversity is crucial in agriculture. It ensures that beneficial species are supported and maintained, while the impact of harmful species is minimized, promoting a more sustainable and productive agricultural system.
In this internship, we aim to develop a framework that utilizes Deep Neural Networks for continual learning in environmental monitoring. In agriculture, the appearance of various species over time necessitates monitoring population dynamics. This critical information allows farmers to make informed decisions regarding their crops. Training such models, especially deep learning models, requires an abundant amount of data. When new species emerge, gathering sufficient data to train a model to detect these newcomers can be challenging due to their novelty in the environment. Our previous research has successfully identified insects in natural settings using advanced algorithms such as You Only Look Once (YOLO) and Region-based Convolutional Neural Networks (R-CNNs). For the development of this model, we have employed cutting-edge techniques, including Active Learning and Transfer Learning. The challenge of this internship is to build on previous work, successfully adapting to new changes and quickly learning to detect newly emerging species.
In contemporary times, there is a growing urgency to enhance the efficiency of production, especially in the agricultural sector, which is the world's primary food supplier. This approach is perfectly aligned with the future trajectory of our society and its evolving standards.
Objetivo
In this internship, the student is expected to study, propose, collect data, implement, and test models to develop a framework for continual learning in intelligent biodiversity detection. To achieve this goal, the following objectives will be pursued:
- Study the state of the art
- Study the available datasets, namely the tomato white fly dataset (Nieuwenhuizen 2018) and the bird net dataset (kaggle) together with datasets form the project associated with this proposal; and/or create new ones
- Study the available frameworks for model development and continual learning
- Define, implement, and fine tune the learning architecture
- Propose and deploy test setup
Plano de Trabalhos - Semestre 1
- Literature review
- Identification and annotation of datasets for model development
- Identification and study of available frameworks for model development
- Define the architecture of the system
- Start implementing the proposed approach
- Write intermediate report
Plano de Trabalhos - Semestre 2
- Implement the proposed solution and fine tune models
- Test and evaluate performance
- Write final report
Condições
This work should take place in the context of a research project funded by PRR. There is the possibility of a 6-month scholarship.
Orientador
Catarina Silva, Dinis Costa, Bernardete Ribeiro
catarina@dei.uc.pt 📩