Machine learning and multiscale methods in the identification of bivalve larvae

Tiwari,; Gallager,

doi:10.1109/iccv.2003.1238388

Cited by 17 publications

(26 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…2). The optical path setup was similar to that used previously Gallager 2003a, 2003b), but omitting bleaching and using a different wave compensation plate prohibited cross-comparisons with the 2003 images. Bleaching shells was not shown to affect classification accuracy (Thompson 2011).…”

Section: Hatchery Rearingmentioning

confidence: 99%

“…Each image from our training sets was run through feature extraction software implemented in MATLAB and identical to that used in Tiwari and Gallager (2003b).…”

Section: Image Preprocessingmentioning

confidence: 99%

“…Cross-polarized light and a full wavelength compensation plate create color patterns that reflect the crystal orientations. These color-patterns are speciesspecific and can be used in pattern-recognition software Gallager 2003a, 2003b). Initial work using six species of preserved hatchery larvae showed accuracies between 80% to 90% (Tiwari and Gallager 2003b).…”

mentioning

confidence: 99%

“…Initial work using six species of preserved hatchery larvae showed accuracies between 80% to 90% (Tiwari and Gallager 2003b). As only color patterns are used as features, polarization techniques are insensitive to shell orientation, size, and morphology Gallager 2003a, 2003b), eliminating many of the ambiguities involved in differentiating bivalve larvae (Perino et al 2008).…”

mentioning

confidence: 99%

“…The similar morphologies of veliger larvae make them less amenable to traditional identification methods using size features and black-and-white images (Hendriks et al 2005), but color images of larvae under polarized light show distinct birefringence patterns Gallager 2003a, 2003b). Once a larva begins shell formation, each species uses a specific protein matrix to control the orientation of the aragonite crystals forming the shell.…”

mentioning

confidence: 99%

See 4 more Smart Citations

Semi‐automated image analysis for the identification of bivalve larvae from a Cape Cod estuary

Thompson

Hare

Gallager

2012

Limnology & Ocean Methods

View full text Add to dashboard Cite

Machine-learning methods for identifying planktonic organisms are becoming well-established. Although similar morphologies among species make traditional image identification methods difficult for larval bivalves, species-specific shell birefringence patterns under polarized light permit identification by color and texturebased features. This approach uses cross-polarized images of bivalve larvae, extracts Gabor and color angle features from each image, and classifies images using a Support Vector Machine. We adapted this method, which was established on hatchery-reared larvae, to identify bivalve larvae from a series of field samples from a Cape Cod estuary in 2009. This method had 98% identification accuracy for four hatchery-reared species. We used a multiplex polymerase chain reaction (PCR) method to confirm field identifications and to compare accuracies to the software classifications. Image classification of larvae collected in the field had lower accuracies than both the classification of hatchery species and PCR-based identification due to error in visually classifying unknown larvae and variability in larval images from the field. A six-species field training set had the best correspondence to our visual classifications with 75% overall agreement and individual species agreements from 63% to 88%. Larval abundance estimates for a time-series of field samples showed good correspondence with visual methods after correction. Overall, this approach represents a cost-and time-saving alternative to molecular-based identifications and can produce sufficient results to address long-term abundance and transport-based questions on a species-specific level, a rarity in studies of bivalve larvae.

show abstract

Section: Hatchery Rearingmentioning

confidence: 99%

“…Each image from our training sets was run through feature extraction software implemented in MATLAB and identical to that used in Tiwari and Gallager (2003b).…”

Section: Image Preprocessingmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations