1 |
Narrow Pretraining of Deep Neural Networks : Exploring Autoencoder Pretraining for Anomaly Detection on Limited Datasets in Non-Natural Image DomainsEriksson, Matilda, Johansson, Astrid January 2022 (has links)
Anomaly detection is the process of detecting samples in a dataset that are atypical or abnormal. Anomaly detection can for example be of great use in an industrial setting, where faults in the manufactured products need to be detected at an early stage. In this setting, the available image data might be from different non-natural domains, such as the depth domain. However, the amount of data available is often limited in these domains. This thesis aims to investigate if a convolutional neural network (CNN) can be trained to perform anomaly detection well on limited datasets in non-natural image domains. The attempted approach is to train the CNN as an autoencoder, in which the CNN is the encoder network. The encoder is then extracted and used as a feature extractor for the anomaly detection task, which is performed using Semantic Pyramid Anomaly Detection (SPADE). The results are then evaluated and analyzed. Two autoencoder models were used in this approach. As the encoder network, one of the models uses a MobileNetV3-Small network that had been pretrained on ImageNet, while the other uses a more basic network, which is a few layers deep and initialized with random weights. Both these networks were trained as regular convolutional autoencoders, as well as variational autoencoders. The results were compared to a MobileNetV3-Small network that had been pretrained on ImageNet, but had not been trained as an autoencoder. The models were tested on six different datasets, all of which contained images from the depth and intensity domains. Three of these datasets additionally contained images from the scatter domain, and for these datasets, the combination of all three domains was tested as well. The main focus was however on the performance in the depth domain. The results show that there is generally an improvement when training the more complex autoencoder on the depth domain. Furthermore, the basic network generally obtains an equivalent result to the more complex network, suggesting that complexity is not necessarily an advantage for this approach. Looking at the different domains, there is no apparent pattern to which domain yields the best performance. This rather seems to depend on the dataset. Lastly, it was found that training the networks as variational autoencoders did generally not improve the performance in the depth domain compared to the regular autoencoders. In summary, an improved anomaly detection was obtained in the depth domain, but for optimal anomaly detection with regard to domain and network, one must look at the individual datasets. / <p>Examensarbetet är utfört vid Institutionen för teknik och naturvetenskap (ITN) vid Tekniska fakulteten, Linköpings universitet</p>
|
2 |
Detective Quantum Efficiency in the Image Domain / Detektiv Kvanteffektivitet i BilddomänenLundhammar, Per January 2022 (has links)
We investigate the signal and noise transform characteristics from the projection domain to the image domain in a linear systems theory framework and predictions of the DQE within this framework are made. A simulation study of a photon counting silicon detector is made from which the energy information is used in order to produce synthetic monoenergetic images. From these images the MTF, NPS, and DQE are estimated and are compared to the respective quantities predicted from the model. Within this model we find that the DQE in the projection domain has similar global characteristics as in the image domain, and that the sampling step in the imaging chain affects the DQE close to the zero frequency. / Vi undersöker hur signal och brus egenskaper transformeras från projektionsdomänen till bilddomänen i ett linjärt-skiftinvariant system och förutsägelser av DQE inom detta ramverk görs. Vidare görs en simuleringsstudie av en fotonräknande kiseldetektor från vilken energiinformationen används för att producera syntetiska monoenergetiska bilder. Från dessa bilder uppskattas MTF, NPS och DQE och jämförs med respektive storheter förutspådda från modellen. Inom modellen finner vi att DQE i projektionsdomänen har liknande globala egenskaper som i bilddomänen, och att samplingssteget i bildkedjan påverkar DQE nära nollfrekvensen.
|
Page generated in 0.0557 seconds