Facial Recognition May Complement Taste Tests

The first exper­i­men­tal method that uses facial recog­ni­tion of emo­tions to com­ple­ment panel tests in the clas­si­fi­ca­tion of extra vir­gin olive oils has been devel­oped in Spain.

Citoliva, Inoleo, and the Emotion Research Lab worked together to design the facial recog­ni­tion soft­ware, which would allow judges to ana­lyze the uncon­scious response of a tast­ing panel to an olive oil sam­ple.

Panel tests focus on the eval­u­a­tion of organolep­tic traits and olive oil defaults. This inno­v­a­tive soft­ware could be used to help tasters ascer­tain how they feel toward olive oil sam­ples in real time.

The use of tech­nol­ogy to ana­lyze feel­ings makes eval­u­a­tion and clas­si­fi­ca­tion of olive oils more objec­tive, pro­vid­ing a greater con­fi­dence to olive oil pro­duc­ers and con­sumers.

The emo­tional organolep­tic test is being devel­oped and tested as part of the OlivEMOTION project.

“Methodology and results of this research project are still pre­lim­i­nary,” Raquel Costales Rodríguez, from Citoliva’s research and devel­op­ment depart­ment, told Olive Oil Times.

Citoliva has been respon­si­ble for gath­er­ing data about emo­tional reac­tions of olive oil tasters using a soft­ware cre­ated with the Emotion Research Lab, which has been adapted for olive oil eval­u­a­tion.

Using either a com­puter or cell­phone cam­era to cap­tures facial reac­tions, the soft­ware has been able to trans­late move­ments and micro-expres­sions of tasters into emo­tional reac­tions that describe the degree of plea­sure or dis­like pro­voked by the olive oils’ dif­fer­ent fla­vors and tex­tures.

“The soft­ware iden­ti­fies pri­mary emo­tions such as hap­pi­ness, sur­prise, anger, dis­gust, fear, sad­ness or neu­tral­ity (the absence of emo­tions), and over a hun­dred sec­ondary emo­tions, such as bore­dom, unease, plea­sure or curios­ity,” Costales Rodríguez said.

The study has ana­lyzed a wide selec­tion of extra vir­gin olive oils that were pre­vi­ously eval­u­ated by Citoliva. The results of both eval­u­a­tions were com­pared in order to repro­duce the tasters’ sen­so­r­ial pro­files of the olive oils. Researchers were then able to clas­sify the oils and inves­ti­gate the con­nec­tion between the facial and ver­bal responses of tasters.

“We ana­lyzed both extra vir­gin olive oils and lam­pante olive oils from dif­fer­ent olive vari­eties and var­i­ous regions,” Costales Rodríguez said. ​“Intensity and defaults were also eval­u­ated. In terms of organolep­tic pro­files, there were oils with dif­fer­ent degrees of fruiti­ness, bit­ter­ness and pun­gency.”

“We tried to secure the great­est pos­si­ble vari­abil­ity with the aim of obtain­ing robust math­e­mat­i­cal mod­els,” she added.

The pre­lim­i­nary con­clu­sions of the results indi­cated that the soft­ware appears to be a suit­able tool to com­ple­ment panel tests in the eval­u­a­tion and clas­si­fi­ca­tion of extra vir­gin olive oils.

Researchers at Citoliva believe that in the future, the tool can be used to reduce inse­cu­rity in the mar­ket­ing of olive oils as well as ensure con­sumer con­fi­dence. The tech­nol­ogy could assuage the fears of export­ing groups who have pre­vi­ously said that results from tast­ing pan­els are too sub­jec­tive and put exporters at legal risk.

According to Costales Rodríguez, Citoliva will con­tinue devel­op­ing the project and based on the results, will deter­mine whether the non-profit will for­mally incor­po­rate this method­ol­ogy to stan­dard assess­ments of olive oils.

• Citoliva


• Emotion Research Lab