AboutZoos, Since 2008


Easy 3D imag­ing of wild ani­mals can sup­port con­ser­va­tion of endan­gered animals

pub­lished 15 July 2018 | mod­i­fied 15 July 2018

On the basis of a few pho­tos alone, a new tech­nique cre­ates real­is­tic avatars of ani­mals that look and move like real ani­mals. This could be help­ful to assess endan­gered ani­mals in the wild as part of species pro­tec­tion projects.

Film­mak­ers and devel­op­ers of com­puter games will have a new way of ani­mat­ing ani­mals in the future. A team led by researchers at the Max Planck Insti­tute (MPI) for Intel­li­gent Sys­tems in Tübin­gen has devel­oped a tech­nique that uses pho­tographs alone to cre­ate life­like 3D mod­els of almost all quadrupeds. These avatars can be ani­mated to real­is­ti­cally imi­tate the move­ments of ani­mals. But the sim­ple method of bring­ing ani­mals to life on the com­puter is not only inter­est­ing for the enter­tain­ment indus­try. It could also ben­e­fit biol­o­gists in species pro­tec­tion and help to make chil­dren in par­tic­u­lar aware of the impor­tance of bio­di­ver­sity. The new tech­nique and its results are pub­lished as part of the Pro­ceed­ings of the IEEE Con­fer­ence on Com­puter Vision and Pat­tern Recog­ni­tion (CVPR), 20 June 2018. And the model is made avail­able for non-​commercial and sci­en­tific research pur­poses (see SMALR).

animals 3d images max planck instituteThe Tübin­gen researchers’ tech­nique uses pho­tos of ani­mals (bot­tom row) to cre­ate 3D mod­els (top row), which can repro­duce dif­fer­ent pos­tures and whose sur­face is designed accord­ing to the orig­i­nal body cover (mid­dle two rows).
© MPI for Intel­li­gent Systems

The key is to start with a rough 3D shape model that can explain the shapes of many ani­mals and then refine this based on the observed images.

Michael J. Black, co-​author, direc­tor Per­ceiv­ing Sys­tems depart­ment, Max Planck Insi­tute for Intel­li­gent Sys­tems, Tübin­gen, Germany.

Pop­u­lat­ing the vir­tual world with ani­mals used to be a lot of man­ual work“, says Black, direc­tor of the Per­ceiv­ing Sys­tems depart­ment at the MPI for Intel­li­gent Sys­tems. Today, ani­ma­tors have to labo­ri­ously design 3D ani­mals using graph­ics soft­ware. While there are tech­nolo­gies to cre­ate real­is­tic 3D avatars of humans, no such tech­nol­ogy is avail­able to auto­mat­i­cally cap­ture the shape of wild ani­mals. “You can’t bring a tiger into the lab to be scanned and you can’t take expen­sive scan­ners out into nature,” says Black.

The key is to start with a rough 3D shape
The team, whose tech­nique cre­ates avatars of ani­mals includ­ing their fur or skin with rea­son­able effort, includes sci­en­tists from the CNR Insti­tute for Applied Math­e­mat­ics and Infor­ma­tion Tech­nol­ogy (IMATI) in Milan and the Uni­ver­sity of Cal­i­for­nia in Berke­ley as well as Max Planck researchers. “Ulti­mately, 3D ani­mal cap­ture is about recon­struct­ing the 3D struc­ture from 2D images of mov­ing objects cap­tured by var­i­ous unknown and mov­ing cam­eras,” explains Sil­via Zuffi, the lead author on the project from IMATI.

In their new method, the researchers only need a few pho­tos of an ani­mal in dif­fer­ent views to cal­cu­late real­is­tic mod­els that can then be ani­mated nat­u­rally. “You could make a for­tune doing this for people’s pets”,” says Michael J. Black.

Unlike other approaches, the images from which the avatars are gen­er­ated can also come from dif­fer­ent cam­eras. The com­puter pro­gramme deliv­ers such con­vinc­ing results because it doesn’t start from scratch when it cre­ates a model. “The key is to start with a rough 3D shape model that can explain the shapes of many ani­mals and then refine this based on the observed images,” says Michael Black.

Help­ing endan­gered ani­mals in the wild
animals 3d image tasmanian tigerTas­man­ian tiger
In their cur­rent work the researchers demon­strate that their method works for as dif­fer­ent ani­mals as a tiger, a bear, a horse or a rhino. “Our algo­rithm gen­er­ates real­is­tic mod­els of ani­mals for which no pre­vi­ous 3D mod­els exist,” says Sil­via Zuffi. For exam­ple, the researchers were able to recon­struct the 3D shape of the extinct Tas­man­ian tiger from low-​quality film footage shot in the 1930’s.

The researchers now want to make the method fully auto­mated so that it can be run on thou­sands of images of ani­mals in the wild. Fully auto­mated 3D ani­mal shape cap­ture would not only facil­i­tate the work of game devel­op­ers and ani­ma­tors but also sup­ports ani­mal con­ser­va­tion. Given pho­tographs of ani­mals cap­tured by cam­era traps, the method could deter­mine whether their shape, i.e. their weight, has changed. Species con­ser­va­tion­ists would thus receive quick and reli­able state­ments as to whether endan­gered ani­mals are doing well.

(Source: Max Planck Gesellschaft news release, 21.06.2018)

UN Biodiversity decade
WWF Stop Wildlife Crime
Fight for Flight campaign
End Ivory-funded Terrorism
Support Rewilding Europe
NASA State of Flux

Goal: 7000 tigers in the wild

Tiger range countries map

Tiger map” (CC BY 2.5) by Sander­son et al., 2006.

about zoos and their mis­sion regard­ing breed­ing endan­gered species, nature con­ser­va­tion, bio­di­ver­sity and edu­ca­tion, which of course relates to the evo­lu­tion of species.
Fol­low me on: