Scene Classification Using a Hybrid Generative/Discriminative Approach

We investigate whether dimensionality reduction using a latent generative model is beneficial for the task of weakly supervised scene classification. In detail, we are given a set of labeled images of scenes (for example, coast, forest, city, river, etc.), and our objective is to classify a new image into one of these categories. Our approach consists of first discovering latent ";topics"; using probabilistic Latent Semantic Analysis (pLSA), a generative model from the statistical text literature here applied to a bag of visual words representation for each image, and subsequently, training a multiway classifier on the topic distribution vector for each image. We compare this approach to that of representing each image by a bag of visual words vector directly and training a multiway classifier on these vectors. To this end, we introduce a novel vocabulary using dense color SIFT descriptors and then investigate the classification performance under changes in the size of the visual vocabulary, the number of latent topics learned, and the type of discriminative classifier used (k-nearest neighbor or SVM). We achieve superior classification performance to recent publications that have used a bag of visual word representation, in all cases, using the authors' own data sets and testing protocols. We also investigate the gain in adding spatial information. We show applications to image retrieval with relevance feedback and to scene classification in videos

Treatment of rats with a self-selected hyperlipidic diet, increases the lipid content of the main adipose tissue sites in a proportion similar to that of the lipids in the rest of organs and tissues

Adipose tissue (AT) is distributed as large differentiated masses, and smaller depots covering vessels, and organs, as well as interspersed within them. The differences between types and size of cells makes AT one of the most disperse and complex organs. Lipid storage is partly shared by other tissues such as muscle and liver. We intended to obtain an approximate estimation of the size of lipid reserves stored outside the main fat depots. Both male and female rats were made overweight by 4-weeks feeding of a cafeteria diet. Total lipid content was analyzed in brain, liver, gastrocnemius muscle, four white AT sites: subcutaneous, perigonadal, retroperitoneal and mesenteric, two brown AT sites (interscapular and perirenal) and in a pool of the rest of organs and tissues (after discarding gut contents). Organ lipid content was estimated and tabulated for each individual rat. Food intake was measured daily. There was a surprisingly high proportion of lipid not accounted for by the main macroscopic AT sites, even when brain, liver and BAT main sites were discounted. Muscle contained about 8% of body lipids, liver 1-1.4%, four white AT sites lipid 28-63% of body lipid, and the rest of the body (including muscle) 38-44%. There was a good correlation between AT lipid and body lipid, but lipid in"other organs" was highly correlated too with body lipid. Brain lipid was not. Irrespective of dietary intake, accumulation of body fat was uniform both for the main lipid storage and handling organs: large masses of AT (but also liver, muscle), as well as in the"rest" of tissues. These storage sites, in specialized (adipose) or not-specialized (liver, muscle) tissues reacted in parallel against a hyperlipidic diet challenge. We postulate that body lipid stores are handled and regulated coordinately, with a more centralized and overall mechanisms than usually assumed.