Search

International audience

International audience; Artificial Intelligence based photography beauty assessment have received a great attention in the last 25 years. They may now claim noticeable performances in replacing the human observer. However, they face limitations which are rooted in the basic choices of the machine learning stage, borrowed from the old Platonism, i.e. the poor place let to a specific observer in the assessment value. Several tracks are explored to short-cut these limitations, based on very different approaches. This paper is presenting an overview of these proposals to re-introduce subjectivity in computational aesthetic assessment and to discuss their foundations.; Les techniques à base d’intelligence artificielle dont l’objectif est d’évaluer automatiquement la qualité esthétique d’une photographie, ont reçu une attention notable ces dernières années et peuvent se targuer de performances prometteuses. On constate cependant que la plupart d’entre-elles souffrent de limitations en raison de leur paradigme de base emprunté à l’esthétique platonicienne, qui attribue tous les critères de beauté à l’objet ou à la personne belle. A partir du très volumineux corpus consacré à l’esthétique depuis 25 siècles, ces limitations auraient pu être anticipées. Le reproche le plus fréquemment exprimé est que le jugement porté sur l’image ne prend pas en compte l’observateur et sa subjectivité. Sans surprise, plusieurs travaux très récents s’attaquent ce point délicat adoptant des approches diverses. Nous les discutons ici.

National audience; On fait le point sur l'état de l'art dans le domaine de la détermination automatique de la beauté d'une photographie

International audience; Under the impulse of machine learning techniques, digital aesthetics assessment received a renewed interestin recent years. In the last 3 years, deep neural networks outclassed hand-crafted feature methods based on image processing and classification. Since then, a handful of studies claim their capacity to separate nice images from the run of the mill production and exhibit scores of almost 80 % agreement with human experts. But what do these methods measure? Implicitly they are based on the "objectivist" tradition of aesthetics dating back to the Greek philosophers, and highly influential on the artistic field up to the 18th century. However, the "subjectivist" point of view, as pioneered by Locke and Burke gained in popularity in the 19th and 20th centuries. Rested on the psychoanalytic school, then by experimental psychology and social studies, and at last in recent days by neuro-biology (and the so called "neuro-aesthetic" trend), the "subjectivist" school gained in support in the scientific community, ... but not in the image-processing and artificial intelligence body! We will show how the history of "scientific beauty evaluation" since the early works of C. Henry (1885) and G. Birkhoff (1933) until DNN is indeed following an identical slope where only few attention is paid to the viewer when most of the literature on aesthetics tells us that other tracks may be more valuable.

—In this paper, we propose a generic multi-temporal SAR despeckling method to extend any single-image speckle reduction algorithm to multi-temporal stacks. Our method, RAtio-BAsed multi-temporal SAR despeckling (RABASAR), is based on ratios and fully exploits a " super-image " (i.e. temporal mean) in the process. The proposed approach can be divided into three steps: 1) calculation of the " super-image " through temporal averaging; 2) denoising the ratio images formed through dividing the noisy images by the " super-image " ; 3) computing denoised images by multiplying the denoised ratio images with the " super-image ". Thanks to the spatial stationarity improvement in the ratio images, denoising these ratio images with a speckle-reduction method is more effective than denoising the original multi-temporal stack. The data volume to be processed is also reduced compared to other methods through the use of the " super-image ". The comparison with several state-of-the-art reference methods shows numerically (peak signal-noise-ratio, structure similarity index) and visually better results both on simulated and real SAR stacks. The proposed ratio-based denoising framework successfully extends single-image SAR denoising methods in order to exploit the temporal information of a time series.

National audience; L'indexation des images satellitaires utilise aujourd'hui abondamment les méthodes classiques d'apprentissage automatique (détection de primitives, classification, etc.) mais pas les réseaux de neurones profonds malgré des besoins notables en sémantique. Pourquoi ?

National audience; On discute des façons de verifier les approches objectiviste (de Platon par exemple) ou subjectiviste (de Locke par exemple) de l'esthétique par des techniques modernes : neuro-biologie ou psychologie expérimentale.

International audience