Welcome to the Conservation Research homepage.
New! A prototype system for monitoring roaming dogs. In order to control the number of free-roaming dogs, animal shelters in many Indian cities are collecting dogs from the city streets to vaccinate, neuter and replace. More CNVR projects are proposed. The amount and type of data recorded is very variable, so potential information about the dogs and the effectiveness of the projects is being lost. The following link describes a way of using a "smartphone" to collect and store the data without investing a lot of time and effort. Dogs of different types (male, female, lactating etc.) seen on the street are recorded by tapping buttons on the screen. Those that are picked up are photographed and described using a voice recording. At the shelter or mobile clinic the data is uploaded to a database. Further information about the dogs and their treatment is added there. Each day the database prints out a list of dogs ready for surgery and sends to the 'phone a map of release locations for dogs ready for release. It also displays monthly reports of the shelter's work and displays maps showing the distribution of dogs of different types and changes over time. Please click straydog database to view it and try it out (click on straydog for other programs relating to roaming dogs)
New! A second system uses the phone purely as a survey tool, to monitor the density of roaming dogs (dogs seen per km) along one or more standard routes over time. The method can be used to monitor the effectiveness of any animal birth control (ABC) intervention because the average number of dogs seen per km gives the expected number that a resident will encounter, for example, while travelling to work or school and hence relates closely to perceived dog "problem". Please click DogDensity to view it and try it out
The "dunsurv" link will allow you to download the "dungsurv" software, which was introduced many years ago at a conference on Asian elephants in Bangalore (hosted by Dr. Raman Sukumar). It allows estimates of animal population density to be derived from surveys of "signs", such as dung, tracks or nests, without making "steady state" assumptions. dungsurv
The "SCANS II" link is for those interested in the aerial survey component of the SCANS II cetacean surveys. From there you can download a simplified simulation program that tests the performance of the "circling" (or "RaceTrack") method used in the surveys. SCANS_II
The remaining links will provide access to the automated photo-id systems we've developed for various species. Over the next few months we will upload more programs and sample image sets that will allow anyone interested in maintaining an automated photo-id catalogue for that species to try using the system and check its suitability for their work.
shark cheetahs tigers seal wildebeest zebra salamander chital sand_lizard crested_newt adders clouded_leopard leopard frog lynx giraffe wilddogs cobras
The follwing animation illustrates extracting a pattern from the flank region of an adult female grey seal via the 3D model (thanks to Wully Patterson, SMRU). Flank_extract_video
NOTE1: on upgrading from Windows XP it is possible that one of the pattern-matching algorithms used in these systems may not run when the ExtractCompare software is installed to the c:\Program Files folder, which is the default location chsoen by the ExtractCompare_install facility. It is recommended to install to another location (not the desktop) or to move the existing ExtractCompare folder from c:\Program Files to another location (remember to edit "path to software" on the database window).
NOTE2: Since January 2011 a free download of software called "Wild ID" has been avialable from: http://software.dartmouth.edu/Macintosh/Academic/Wild-ID_1.0.0.zip. The WIld ID software is a highly effective implementation of the SIFT algorithm for matching animals patterns and depending on the particular pattern and range of camera angle and animal posture in the available images the Wild ID program may be sufficient for identification without the need to scan the pattern via a 3D model, as in the links above.
Lex Hiby, 21st November 2012