56. Seek (offline database, no wifi) Google Lens, PlantNet, Flora Incognita, Plant.ID

Testen van diverse identificaties apps. Wel jammer dat dan weer Seek gebruikt wordt ipv de iets betere iNaturalist.org app. Seek heeft al bijzondere feature dat het offline gebruikt kan worden. Het zou wat zijn als Seek verder als offline app gemaakt zou worden..

https://bsbi.org/wp-content/uploads/dlm_uploads/BSBI-News-144-pp34-40-plant-id-apps-final.pdf
https://entomologytoday.org/2023/10/31/growing-utility-online-photo-sharing-entomology-research/
https://arxiv.org/pdf/2203.03253v1.pdf
https://inf.news/en/tech/fe120ee6c96f52c780e3a562e6de7f88.html
https://www.cambridge.org/core/journals/lichenologist/article/can-we-trust-inaturalist-in-lichenology-evaluating-the-effectiveness-and-reliability-of-artificial-intelligence-in-lichen-identification/B3FF5A93621A4684B5E6A114F7C89FFB
https://floraincognita.com/publications/
https://floraincognita.com/blog/2023/10/04/adopting-flora-incognita-plant-observations-for-phenology-monitoring/

Seek (cf iNaturalist, offline app, no WiFi required)

This app uses the iNaturalist database and is very different from the others tested; all the user needs to do is point the camera at the object plant and the app performs a real-time evaluation from the live video feed, often rapidly improving through class, order, family, genus and even to species as the view is changed. Secondly, and critically, it is the only app that does not require internet access to operate. Although this was the weakest app for monocots in this test, it was only just beaten by Google Lens in terms of the overall percentage of ids correct to family or genus, while it was outstanding in the fact that it was very conservative and made by far the fewest wrong ids of any app tested (where wrong ids were assessed as the wrong genus or family).

It is notable that Seek, although it does not have a particularly high rate of identification to species level, is among the best at identifying to genus and to family, but more importantly it is generally conservative, only making an identification to a level with which it is confident, so that it has the lowest error rate of all apps tested.

Google Lens

This app is widely available both for Android (from Play Store) or as a component of Google photos for iOS. Overall it is much more wide-ranging than a plant identification app, as it will attempt to identify almost anything. Nevertheless, it performed extremely well on our test set of plant images though with a tendency to identify plants as North American species rather than the correct British species; this led to it only ranking third in terms of the percentage ids correct to species, though it was second-best in the ids correct to genus and to family. Disadvantages of Google Lens are that there is no option for feedback to
orrect errors and it only gives a broad hiera

Flora Incognita

This was also an impressive app, achieving the second highest rate of identification to species level
of all apps tested and good rates of identification to genus and to family. To use this app the user
first needs to make an initial identification to herb/ shrub, tree, grass/sedge or fern. The app then gives
the user options of taking photos of leaves, flowers, fruits or the whole plant. The app then provides an
identification when it has enough information. For the present tests, to allow direct comparability with
the other apps, only a single image was provided each time. It is likely that the use of additional photos
would improve the accuracy of this app furthe

Andere Apps

Results for the five other free apps tested (Table 1) were omitted as they did not match those in Table 2.
Even the best of them, PlantSnap, identified fewer than 50% of samples correctly to family, but more
seriously, 61% of its first suggestions were the wrong genus or family. I therefore doubt that this would
be very useful for beginners. Candide is aimed at gardeners and although not good on the set of images
used here (which were primarily wild species), it may be a good choice for identification of garden plants
and other aliens.

Seek (iNaturalist.org)

It is notable that Seek, although it does not have a particularly high rate of identification to species
level, is among the best at identifying to genus and to family, but more importantly it is generally
conservative, only making an identification to a level with which it is confident, so that it has the lowest
error rate of all apps tested. https://bsbi.org/wp-content/uploads/dlm_uploads/BSBI-News-144-pp34-40-plant-id-apps-final.pdf

iNaturalist currently uses vision models in two main places:
1) a private web-based API used by the website and the iNaturalist iOS and Android apps, and
2) within the recently updated Seek app.

When Seek 2.0 was released in April, it included a different vision model than we were using on the web. At that time the web-based model was a third-generation model we started using in early 2018. That web-based model was trained with the idea it would be run on servers, and servers can be configured to have far more computing power than a mobile device. As a result that model was far too large to be run on mobile devices.

Early this year, with an updated Seek in mind, we started another training run with two main goals:
-shrinking the file size of the model, and
-allowing it to recommend taxonomic ranks other than species (e.g. families, genera, etc.).

Smaller
The mobile version of the model needs to be small in terms of file size to minimize the amount of data app users would need to download. Smaller models can also be used by more devices as they need fewer resources to run (e.g. memory, battery), and can generate results faster, which is important for Seek's real-time camera vision results. These models take a lot of time and money to train, so we also wanted a model that could be simultaneously trained to produce a large web-based version and a smaller version for use in mobile devices.

Unfortunately, shrinking the file size like this slightly decreased model accuracy compared to the larger web-based version (kind of similar to image compression), and we found that was an unavoidable tradeoff. We take this into account when processing the model results, and on average for a similar error rate, the mobile version might recommend a taxon at a higher taxonomic rank than the web-based version. The taxon results we show to users shouldn't be less accurate, but they may be less specific.

More Species Represented
We wanted the model to include more species data, even when some species don't have enough photos to be recognized as species level. There are some species with a small amount of photos that, if we trained on that small set of photos, likely wouldn't have enough information for the model to reliable recognize those species.

Our 2018 model only included taxa at rank species. We set a threshold for number of photos, and species below the threshold were not included. We could still recommend higher taxa by doing some post-processing of results, but the model itself would only assign scores to species. In our latest training run we allowed the photos from species under the threshold to be rolled up into their ancestor taxa until the threshold was reached, and we allowed the model to assign scores to these non-species nodes. This allows more species to be represented in this newer model, sometimes at the genus level mixed up with photos of other species in the genus under our threshold. Now instead of not knowing anything about these species, the model can at least identify the genus or family, etc.

https://www.inaturalist.org/blog/25510-vision-model-updates
https://bsbi.org/wp-content/uploads/dlm_uploads/BSBI-News-144-pp34-40-plant-id-apps-final.pdf
https://www.inaturalist.org/posts/38046-fotoherkenning-paddenstoelen-een-vloek-of-een-zegen

https://bsbi.org/wp-content/uploads/dlm_uploads/BSBI-News-144-pp34-40-plant-id-apps-final.pdf

https://www.inaturalist.org/posts/38046-fotoherkenning-paddenstoelen-een-vloek-of-een-zegen

Posted on December 27, 2020 06:12 PM by ahospers