Here is an unusual way to sift through the ever-growing number of files stored on your PC 鈥� use your sense of smell.
A team led by Stephen Brewster at the University of Glasgow in Scotland, UK, has created software and hardware to let computer users attach distinctive smells 鈥� the scent of fresh-cut grass or the whiff of chocolate 鈥� to a batch of images. The system, dubbed Olfoto, then lets users sort through their image collection simply by following their noses.
Brewster and colleagues tested volunteers鈥� ability to search through their own digital photos by smell. Each user first attached smells to different images, and then tried searching for a particular snapshot by selecting its odour.
Advertisement
The researchers looked at how well the participants could sniff out their images compared with sorting by conventional text tags. Although sorting by smell was far slower, the results were still encouraging, Brewster says.
Smelly cubes
鈥淚t鈥檚 not perfect, but people can do it,鈥� he told New 杏吧原创. 鈥淚t鈥檚 quite surprising and it鈥檚 under-exploited.鈥�
For their experiments, the researchers used an array of cube-shaped capsules, each of which releases a different fragrance when activated electronically and which can be identified using a radio frequency identification (RFID) reader. When a particular image appeared on-screen a users could 鈥渢ag鈥� it with a scent by moving their chosen cube over the wireless reader.
The ability to search by smell is perhaps not surprising, Brewster says, when you consider the way a smell can evoke powerful memories. 鈥淪mell has interesting possibilities in terms of helping you retrieve memories,鈥� Brewster says. 鈥淎nd, if you have 10,000 images on your computer stretching back 10 years, how do you search through them all?鈥�
The smells used in the study range from flowers and grass to 鈥渟weaty feet鈥� and even the distinct aroma of a farmyard.
Useful smells
Brewster notes that most people can identify thousands of different odours and believes they should get better at searching by smell with practice. However, he says generating smells reliably and delivering them effectively remains a serious problem. 鈥淵ou just can鈥檛 synthesise smells from scratch,鈥� he says. 鈥淲e鈥檙e also currently trying to find the most useful set of smells for this.鈥�
The researchers also found that the smells produced by their cubes changed very slightly as the experiment went on.
Stephen Boyd-Davis, an expert on human-computer interaction at Middlesex University in the UK, says the work might seem odd but has potential. 鈥淚t is an interesting and under-researched area,鈥� he told New 杏吧原创. 鈥淧eople are certainly affected by odours.鈥�
Boyd-Davis agrees that generating and delivering smells is tricky. He also says the Olfoto study fails to distinguish sufficiently between the way a smell simply grabs someone鈥檚 attention and the way it can be used to identify something in particular.
Brewster also concedes that the idea is speculative and will be considered wacky by some. Nevertheless, he plans to push ahead with further smell-related research. He is working on another project involving smells, but is unwilling to reveal the details.
