Unsupervised pre-training was a critical technique for training deep neural networks years ago. With sufficient labeled data and modern training techniques, it is possible to train very deep neural networks from scratch in a purely supervised manner nowadays. However, unlabeled data is easier to obtain and usually of very large scale. How to make use of them better to help supervised learning i...

A novel single face image Super Resolution (SR) framework based on adaptive-block Principal Component Analysis (PCA) is presented in this paper. The basic idea is the reconstruction of a High Resolution (HR) face image from a Low Resolution (LR) observation based on a set of HR and LR training image pairs. The HR image block is generated in the proposed method by using the same position image b...

Generating natural language descriptions of images is an important capability for a robot or other visual-intelligence driven AI agent that may need to communicate with human users about what it is seeing. Such image captioning methods are typically trained by maximising the likelihood of ground-truth annotated caption given the image. While simple and easy to implement, this approach does not ...

The use of image transformations is essential for efficient modeling and learning of visual data. But the class of relevant transformations is large: affine transformations, projective transformations, elastic deformations, ... the list goes on. Therefore, learning these transformations, rather than hand coding them, is of great conceptual interest. To the best of our knowledge, all the related...

Due to small training sets, statistical shape models constrain often too much the deformation in medical image segmentation. Hence, an artificial enlargement of the training set has been proposed as a solution for the problem. In this paper, the error sources in the statistical shape model based segmentation were analyzed and the optimization processes were improved. The method was evaluated wi...

We are concerned with the problem of image segmentation in which each pixel is assigned to one of a predeened nite number of classes. In Bayesian image analysis, this requires fusing together local predictions for the class labels with a prior model of segmentations. Markov Random Fields (MRFs) have been used to incorporate some of this prior knowledge, but this not entirely satisfactory as inf...

People are typically poor at matching the identity of unfamiliar faces from photographs. This observation has broad implications for face matching in operational settings (e.g., border control). Here, we report significant improvements in face matching ability following feedback training. In Experiment 1, we show cumulative improvement in performance on a standard test of face matching ability ...

Cellular automata can be significantly applied in image processing tasks. In this paper, a novel method to train two dimensional cellular automata for detection of edges in digital images has been proposed and experiments have been carried out for the same. Training of two dimensional cellular automata means selecting the optimum rule set from the given set of rules to perform a particular task...

We propose a learning-based image feature points detector. Instead of giving an explicit definition for feature point we apply the methods of machine learning to infer it inductively using a representative training set. This allows for a flexible tuning of the proposed detector to a specific problem that is described by a training set of desired responses. To increase feature points' repeatabil...

For training we have 7049 grayscale images of equal size (96x96). For each training image, 15 keypoints are provided with both x and y coordinates, allowed omission for portion keypoints. Hence mostly there are 30 or fewer target labels for each training image. Test data set has 1783 images and we need to predict the keypoint coordinates for the test images to obtain the Kaggle score (using RMSE).