In this paper, we propose convolutional neural networks for semantic segmentation on road markings in the situation where sequential segmentation ground truth masks are available. The proposed model aggregates the temporal information and the context information from the multiple frames. Moreover, we employ CGNet as the backbone network to reduce trainable parameters and computation speed. In the experiment, we evaluate the model using the Gifu-city Road Marking Segmentation Dataset, which includes road markings of open roads in Gifu city. As a result, the segmentation performance such as a white center line and white dash line is an improvement.
The inspection of solder joints on printed circuit boards is a difficult task because defects inside the joints cannot be observed directly. In addition, because anomalous samples are rarely obtained in a general anomaly detection situation, many methods use only normal samples in the learning phase. However, sometimes a small number of anomalous samples are available for learning. We propose a method to improve performance using a small number of anomalous samples for training in such situations. Specifically, our proposal is an anomaly detection method using an adversarial autoencoder (AAE) and Hotelling’s T-squared distribution. First, the AAE learns features of the solder joint following the standard Gaussian distribution from a large number of normal samples and a small number of anomalous samples. Then, the anomaly score of a solder joint is calculated by Hotelling’s T-squared method from the features learned by the AAE. Finally, anomaly detection is performed by thresholding using this anomaly score. In experiments, we show that our method performs anomaly detection with few false positives in such situations. Moreover, we confirmed that our method outperforms the conventional method using handcrafted features and a one-class support vector machine.
Designing the optimal architecture of neural networks is an important issue. However, since this is difficult even for experienced experts, automatic optimization of the network architecture is required. In this study, we regard this issue as a combinatorial optimization problem, and utilize genetic algorithm to optimize the network architecture. Because training the networks, which are represented by individuals in GA, takes a long time, a novel method to reduce the training time by inheriting the weights of the trained network is proposed. From experimental results, our proposed method achieved the time reduction and higher accuracy than a conventional method.
We propose a defect detection method of solders on a printed circuit board using X-ray CT inspection system and Adversarial Autoencoder (AAE) . We obtain sliced images of the solder using X-ray CT and extract their features that follow the standard normal distribution by using AAE. Then, the solder defects are detected by Hotelling's T square. As a result of experiments, we show that we can classify normal and anomalous data samples completely on the condition of training with large normal samples and small anomalous samples.
In this paper, we aimed at discrimination of defects under conditions where there is a large number of good products and a small number of defective products. Although automation of a visual inspection is essential to improve the quality of products, either or both of the features extracted by the experts and balanced dataset are needed. We tackled such a problem. By combining AAE, which can extract features following any distribution and Hotelling's T-Square, which is an effective anomaly detection method when data follows a normal distribution, it is possible to discriminate defects with a small number of defective samples.
We have been developing the CAD scheme for head and abdominal injuries for emergency medical care. In this work, we
have developed an automated method to detect typical head injuries, rupture or strokes of brain. Extradural and subdural
hematoma region were detected by comparing technique after the brain areas were registered using warping. We employ
5 normal and 15 stroke cases to estimate the performance after creating the brain model with 50 normal cases. Some of
the hematoma regions were detected correctly in all of the stroke cases with no false positive findings on normal cases.