Detecting actions of fruit flies

In this thesis we describe a system that tracks fruit flies in video and automatically detects and classifies their actions. We introduce Caltech Fly-vs-Fly Interactions, a new dataset that contains hours of video showing pairs of fruit flies engaging in social interactions, and is published with complete expert annotations and articulated pose trajectory features. We compare experimentally the value of a frame-level feature representation with the more elaborate notion of bout features that capture the structure within actions. Similarly, we compare a simple sliding window classifier architecture with a more sophisticated structured output architecture, and find that window based detectors outperform the much slower structured counterparts, and approach human performance. In addition we test the top performing detector on the CRIM13 mouse dataset, finding that it matches the performance of the best published method.

Fast growing mode of stimulated Raman scattering in a pure three-wave process

Stimulated Raman scattering (SRS) of a relativistic laser in plasmas is studied in the framework of the standard equation set of a three-wave process. As far as every wave involved in the process is concerned, its evolution has two aspects: time-dependent amplitude and time-dependent frequency. These two aspects affect each other. Strict analysis and numerical experiment on the full three-wave equation set reveal that a fast growing mode of the instability, which could reach a balance or saturation point during a period far shorter than an estimation based on conventional analysis, could take place in a standard three-wave process without coupling with a fourth wave. This fast growing mode is found to stem from the constraint set by the background density on the amplitude of the driven Langmuir wave. The effect of various parameters on the development of the SRS instability is studied by numerical calculation of the history of the instability in different cases. (c) 2007 American Institute of Physics.

The production of algae growing on gravel in a chalk-stream

Epilithic algae, ie that growing on the surface of stones, was studied as part of the work on the energy flow of the chalk-stream ecosystem, by the River Laboratory. The study area was on Bere Stream and 2 neighbouring streams. The algal biomass was estimated from analysis of chlorophyll a. In Bere Stream the peak chlorophyll a cover occurred in April, while in the neighbouring streams, which have considerably lower nutrient levels, there was on peak. Assuming that 2% of a diatoms dry wt is chlorophyll a, then even in mid-April the biomass of epilithic algae amounted to no more than 15 g dry wt m Super(-2) of exposed gravel. Annual production was calculated to be > 15 times greater than biomass. The estimation of net primary production is always difficult for benthic floras and comparisons are especially difficult when different methods are used. But these figures contrast sharply with those for Ranunculus (water crowfoot) which has a ratio of annual production to maximal seasonal biomass of 1:16. The accumulation of algal biomass is apparently being prevented. Some organic matter may be excreted; some algae will be washed off the bed of the stream by current and grazing by herbivorous invertebrates will also tend to prevent algal accumulation.