Paper Reading #2: Exploring Interfaces to Botanical Species Classification

Title:
Paper Reading #2: Exploring Interfaces to Botanical Species Classification

Comments:
Cindy Skach
Joshua Penick

Reference:
White, Sean and Feiner, Steven. Exploring Interfaces to Botanical Species Classification. CHI 2010: Media Showcase Session 3.

Summary:

The goal of White and Feiner's research was to compare different devices in obtaining, comparing and identifying botanical species.  Currently it is difficult to identify plants because of the data collection and expert knowledge needed in order to identify the plants.  The devices used were a Tablet PC, an ultramobile personal computer (UMPC), an Apple iPhone, augmented reality, and Microsoft Surface.
The steps in which were followed for the plant identification were to acquire a plant specimen, identify matching specimen, compare potential matches, inspect details and characteristics of those potential matches, collect contextual data, and send the collected data in a format viable for research.  The comparison between the five devices showed that

1. Tablet PCs, which would require a WIFI camera and GPS receiver in addition to the tablet was useful because of its large screen, however there was noticed to be too many parts to potentially lose and the equipment was to heavy.
2. The UMPC had an integrated camera, therefore only the GPS receiver was needed.  This made the system more mobile.
3. The iPhone had an integrated camera and GPS making it even more mobile than the UMPC.  It required network connection though.
4. The augmented reality tested 2 systems: a pair of glasses that overlaid the virtual plants directly, and a pair of glasses that used a clipboard on which to display the virtual plants.  The results showed that the augmented reality made for easier comparison as one could compare directly side by side in a more natural way than having to switch from looking at a screen to the real thing.
5. The Microsoft Surface was not able to be used for field work, and also was not able to get the detail that the other cameras did.

Discussion:

I found it interesting that the authors seemed to attribute such success to the augmented reality equipment.  My thought is that they are bias towards it because my last paper was by Feiner and one of his augmented reality devices at the same conference.  That said, I really appreciated the extent to which they compared the five interfaces.  The Microsoft Surface was doomed for failure, and I am not really sure why they included that in their study; maybe to prove the Microsoft product as lacking.  I am interested to know why the iPhone needed the algorithm on a separate server.  My guess is because the database was too vast to fit in its storage.

Expansion on this research could be the quality of recognition, the algorithm behind correctly identifying the plant.  That wasn't the point of this paper, but I would be interested in statistics of how well this software was able to work and if it would be able to replace the expert plant identifier.