ISLAB/CAISR - User contributions [en]

Piglets Detection and Counting using Deep Neural Networks

2018-10-11T09:07:02Z

Peter:

{{StudentProjectTemplate
|Summary=Piglets Detection and Counting using Deep Neural Networks
|TimeFrame=Winter 2018, Spring 2019
|References=Yolo: https://pjreddie.com/darknet/yolo/
|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms.
|Supervisor=Peter Berck, Sepideh Pashami,
|Level=Master
|Status=Open
}}
Machine learning is being applied to all areas in society nowadays, and farming is no exception. Another are of deep learning which has gained a huge popularity the last years is image classification and object recognition. Better hardware and the availability of data has made the field of image processing the poster child for deep learning techniques.

This project combines smart farming with object recognition and. Its aim is to detect and count piglets from camera images, using semi-supervised learning and transfer learning. The work will involve finding appropriately labelled training data, training a detection network, and applying the trained network to the unlabelled data from the webcams.

This project needs to be done on-site at Sony in Lund (at least one day a week).

Resolving Class Imbalance using Generative Adversarial Networks

2018-10-11T09:06:44Z

Peter:

{{StudentProjectTemplate
|Summary=Resolving Class Imbalance using Generative Adversarial Networks
|Keywords=GAN, neural networks, deep learning
|TimeFrame=Winter 2018, Spring 2019
|References=NIPS 2016 Tutorial on GANs
https://arxiv.org/pdf/1701.00160.pdf

Effective data generation for imbalanced learning using Conditional Generative Adversarial Networks
https://www.researchgate.net/publication/319672232_Effective_data_generation_for_imbalanced_learning_using_Conditional_Generative_Adversarial_Networks

BAGAN: Data Augmentation with Balancing GAN
https://arxiv.org/abs/1803.09655

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
https://arxiv.org/pdf/1606.03657.pdf
|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms.
|Supervisor=Sepideh Pashami, Peter Berck
|Level=Master
|Status=Open
}}
Huge class imbalance can be a challenge for learning tasks. In such cases, different undersampling and oversampling techniques have been usually used to balance the dataset and compensate for the number of samples in minority class. In this thesis we would like to leverage Generative Adversarial Networks (GANs) for addressing the imbalances in data for classification tasks.

The student expect to perform literature review on GANs specifically different structures such as InfoGAN, Conditional GAN, Auxiliary classifier GANs. They should be able to implement and compare performances of different algorithms using different datasets. The student should be able to compare the results with other undersample and oversampling methods, e.g. SMOTE. Modification of standard approaches for classification task is encouraged. Finally, students should reflect on potential failures and limitations of the methods.

Anomaly Detection on Truck Histograms

2018-10-11T09:06:16Z

Peter:

{{StudentProjectTemplate
|Summary=Anomaly Detection on Truck Histograms
|TimeFrame=Winter 2018, Spring 2019
|References=Learning Low-Dimensional Representation of Bivariate Histogram Data
https://ieeexplore.ieee.org/abstract/document/8464276
|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms
|Supervisor=Sepideh Pashami, Peter Berck
|Level=Master
|Status=Open
}}
There are many bivariate histograms collected on-board of Volvo Trucks which accumulate information about the operation of the vehicles through time. Measuring the number of hours that a vehicle has a certain speed versus torque could be one example of these histograms.

Early detection of anomalous behaviour could be important for improving safety and uptime. Anomalies behaviour can be caused by wear, failure or malfunctions of components in a vehicle. It can also be caused by usage pattern through life time of a vehicle.

This thesis will investigate if a vehicle has been used as usual or in anomalous way by analysing bivariate histogram data. The comparison can be done based on history of an individual vehicle or group of vehicles.

Anomaly Detection on Truck Histograms

2018-10-11T08:07:06Z

Peter: Anomaly Detection on Truck Histograms

{{StudentProjectTemplate
|Summary=Anomaly Detection on Truck Histograms
|TimeFrame=Winter 2018, Spring 2019
|References=Learning Low-Dimensional Representation of Bivariate Histogram Data
https://ieeexplore.ieee.org/abstract/document/8464276

|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms
|Supervisor=Sepideh Pashami, Peter Berck
}}
There are many bivariate histograms collected on-board of Volvo Trucks which accumulate information about the operation of the vehicles through time. Measuring the number of hours that a vehicle has a certain speed versus torque could be one example of these histograms.

Early detection of anomalous behaviour could be important for improving safety and uptime. Anomalies behaviour can be caused by wear, failure or malfunctions of components in a vehicle. It can also be caused by usage pattern through life time of a vehicle.

This thesis will investigate if a vehicle has been used as usual or in anomalous way by analysing bivariate histogram data. The comparison can be done based on history of an individual vehicle or group of vehicles.

Piglets Detection and Counting using Deep Neural Networks

2018-10-10T14:57:03Z

Peter: Created page with "{{StudentProjectTemplate |Summary=Piglets Detection and Counting using Deep Neural Networks |TimeFrame=Winter 2018, Spring 2019 |References=Yolo: https://pjreddie.com/darknet/..."

{{StudentProjectTemplate
|Summary=Piglets Detection and Counting using Deep Neural Networks
|TimeFrame=Winter 2018, Spring 2019
|References=Yolo: https://pjreddie.com/darknet/yolo/

|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms.
|Supervisor=Peter Berck, Sepideh Pashami,
}}
Machine learning is being applied to all areas in society nowadays, and farming is no exception. Another are of deep learning which has gained a huge popularity the last years is image classification and object recognition. Better hardware and the availability of data has made the field of image processing the poster child for deep learning techniques.

This project combines smart farming with object recognition and. Its aim is to detect and count piglets from camera images, using semi-supervised learning and transfer learning. The work will involve finding appropriately labelled training data, training a detection network, and applying the trained network to the unlabelled data from the webcams.

This project needs to be done on-site at Sony in Lund (at least one day a week).

Resolving Class Imbalance using Generative Adversarial Networks

2018-10-10T14:54:26Z

Peter: Created page with "{{StudentProjectTemplate |Summary=Resolving Class Imbalance using Generative Adversarial Networks |Keywords=GAN, neural networks, deep learning |TimeFrame=Winter 2018, Spring ..."

{{StudentProjectTemplate
|Summary=Resolving Class Imbalance using Generative Adversarial Networks
|Keywords=GAN, neural networks, deep learning
|TimeFrame=Winter 2018, Spring 2019
|References=NIPS 2016 Tutorial on GANs
https://arxiv.org/pdf/1701.00160.pdf

Effective data generation for imbalanced learning using Conditional Generative Adversarial Networks
https://www.researchgate.net/publication/319672232_Effective_data_generation_for_imbalanced_learning_using_Conditional_Generative_Adversarial_Networks

BAGAN: Data Augmentation with Balancing GAN
https://arxiv.org/abs/1803.09655

InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets
https://arxiv.org/pdf/1606.03657.pdf

|Prerequisites=Artificial Intelligence and Learning Systems courses; good knowledge of machine learning and neural networks; python programming skills for implementing machine learning algorithms.
|Supervisor=Sepideh Pashami, Peter Berck
}}
Huge class imbalance can be a challenge for learning tasks. In such cases, different undersampling and oversampling techniques have been usually used to balance the dataset and compensate for the number of samples in minority class. In this thesis we would like to leverage Generative Adversarial Networks (GANs) for addressing the imbalances in data for classification tasks.

The student expect to perform literature review on GANs specifically different structures such as InfoGAN, Conditional GAN, Auxiliary classifier GANs. They should be able to implement and compare performances of different algorithms using different datasets. The student should be able to compare the results with other undersample and oversampling methods, e.g. SMOTE. Modification of standard approaches for classification task is encouraged. Finally, students should reflect on potential failures and limitations of the methods.