Image-generating machine learning models are typically trained with loss functions based on distance in the image space. This often leads to over-smoothed results. We propose a class of loss functions, which we call deep perceptual similarity metrics (DeePSiM), that mitigate this problem. Instead of computing distances in the image space, we compute distances between image features extracted by...

Computer Assisted Visual Interactive Recognition (CAVIAR) draws on sequential pattern recognition, image database, expert systems, pen computing, and digital camera technology. It is designed to recognize wild flowers and other families of similar objects more accurately than machine vision and faster than most laypersons. The novelty of the approach is that human perceptual ability is exploite...

(Summary of Results) Abstract In this paper, we propose parallel implementations for solving Perceptual Grouping tasks on distributed memory machines. Our implementations show that, given 7K line segments extracted from a 1K 1K image, the Line Grouping task can be performed in 0.486 seconds using a partition of CM-5 having 256 processing nodes and in 0.382 seconds using a 16-node Cray T3D. The ...

INTRODUCTION Mass-produced user interfaces are inadequate for today’s computing needs: interfaces shipped with today’s complex applications are designed in a “one size fits all” manner and by aiming to address the needs of the average user they miss essential needs of most individual users. In particular, I note that: • users own diverse display-equipped devices, which support different interac...

In this paper we present an implementation of a perceptual completion model performed in the three dimensional space of position and orientation of level lines of an image. We show that the space is equipped with a natural subriemannian metric. This model allows to perform disocclusion representing both the occluding and occluded objects simultaneously in the space. The completion is accomplish...

A new adaptive image watermarking algorithm is proposed in this paper. A perceptual model based on the computation of NVF (noise visibility function) is exploited in the proposed watermarking scheme. The original image is first transformed in the wavelet domain and the embedding factor is decided through computing the NVF of the mid-frequency; then the watermark is embedded in the selected subb...

Perceptual and control systems are tasked with the challenge of accurately and efficiently estimating the dynamic states of objects in the environment. To properly account for uncertainty, it is necessary to maintain a dynamical belief state representation rather than a single state vector. In this review, canonical algorithms for computing and updating belief states in robotic applications are...

A global inversion procedure from the acoustic signal to motor commands is presented here based on a postural target invariance hypothesis. Using a model of vowel production, dynamic motor commands were inferred for a vowel sequence pronounced under different levels of emphasis stress and rate. The results enable to assign a prosodic role to the dynamic parameters of the model and thus to discr...

The doubly circular relations of the major and minor keys based on all twelve pitchclasses can be depicted in toroidal models. We demonstrate a convergence of derivations from the different bases of conventional harmonic theory and recent experiments in music psychology. We present a formalization of the music-theoretical derivation from Gottfried Weber’s 1817 chart of tone-centers by using a t...

Pair comparison methods based on Case V of Thurstone’s Law of Comparative Judgment are widely used to derive interval scales for perceptual image quality. A thorough treatment of the involved statistical errors is often neglected, even though this is the base for computing confidence intervals and other statistical tests. In this paper we show, that consequent error estimation through all steps...