Insect image search engine on the Internet

M. B. are there any extra spiders?

Oops, there are even sunflowers, wow insect! Maybe we should narrow it down.

This post was edited by ENO - 05/23/2016 03: 38

I tried to break through all three of our polists in turn, unlike Yandex and Google, the vast majority of photos (and maybe all of them, I didn't study each one) were determined correctly, which is a big plus. I wish further development to the resource!

M. B. are there any extra spiders?
Oops, there are even sunflowers, wow insect! Maybe we should narrow it down.

People from other regions also asked us to do it in some form, even without a tree, to see how it can be:--) If this remains, then we will need to add an additional parameter (they say, look only for insects or something like that) for greater convenience.
But more there now on insects kanesh decent information...

I tried to break through all three of our polists in turn, unlike Yandex and Google, the vast majority of photos (and maybe all of them, I didn't study each one) were determined correctly, which is a big plus. I wish further development to the resource!

Nice to hear, thank you!

It is better to abandon this idea right away - there will only be confusion and no use.

People are always different :-- ) I myself am also not a fan of local names, but there are some who can't do otherwise. There should be no confusion with the Russian language — bo is Cyrillic (although there may be confusion with the names themselves, but those who search in Russian will survive), with English, on the one hand, it is more complicated, bo is also Latin, but on the other — it is more necessary. As far as I know, they are more often used locally and they are more ordered by tama...

PS. I overlooked it, there was an error on the hosting, and the site was lying for a while. Who got to this period-come in now, everything works.

I tried it. In principle, not bad.
But about an hour ago, the service hung up tightly. Was there an influx of testers?
It doesn't search for everything based on Russian names. For example, the wine hawk moth and pigeon Icarus did not show. Along the way, I found out a strange glitch: I decided to correct the first letter to a capital one, suddenly it will find it so. I usually do it this way: I put the cursor at the position "after the first letter", backspace, and correct it. So, for some reason, the cursor stubbornly did not want to stay there, jumped to the end of the line. I played around with this procedure, it seems to work sometimes, but you need not move the mouse after setting the cursor.

My cursor is behaving appropriately right now. Changing the case, as far as I understand, does not affect the search.

Vorona, what kind of wine hawk moth did you need? There are already three of them in different "standard sizes". And this is one of the many reasons why Russian names are evil at this point in the history of zoological nomenclature. English is even worse.

I was just looking at how to search for Russian names. Something is looking for, something is not. Why — I don't know, maybe it's really because of the ambiguity.

But about an hour ago, the service hung up tightly. Was there an influx of testers?

Good question. If there are regular problems with the same request — please reset the link and I'll see if it's the problem. This is if it is tight.
If not, it's probably just a short-term glitch on the hosting service.

It doesn't search for everything based on Russian names. For example, the wine hawk moth and pigeon Icarus did not show.

In the future, if you bother with Russian names, you will need to search for various combinations of words. Right now, it only searches for one variant (i.e., the search engine doesn't know that wine hawk moth and wine hawk moth are the same thing). For example, you can find a wine hawk moth by searching for "hawk moth", "hawk moth b", "wine hawk moth", but you can't find it by searching for "wine hawk moth" or "wine hawk moth":
http://ist.insecta.pro/?cap=%D0%B1%D1%80%D...%B8%D0%BD%D0%BD
Golubyanka ikar is also being sought:
http://ist.insecta.pro/?cap=голубянка+икар

Plus, I have now removed the restriction on the exact occurrence. That is, until recently, only "hawk moth wine" was searched, and "hawk moth winn" was not searched. Now we'll find him.

In addition, adding local names is not the highest priority, so they are still only available in the database for a few thousand taxa (for comparison, photos of about two hundred thousand taxa are currently indexed).

Along the way, I found out a strange glitch: I decided to correct the first letter to a capital one, suddenly it will find it so. I usually do it this way: I put the cursor at the position "after the first letter", backspace, and correct it. So, for some reason, the cursor stubbornly did not want to stay there, jumped to the end of the line. I played around with this procedure, it seems to work sometimes, but you need not move the mouse after setting the cursor.

Wow. If you can simulate this again, please send us a link to the page where this happens and the name of the browser you are using.

This post was edited by Asar - 05/23/2016 23: 45

Gentlemen who vote for "bad", once again I ask you to voice what exactly is bad, so that your vote has not only general informational significance, but also contributes to correcting this very bad.

Cursor dances. It turns out everywhere
I didn't add that after setting the cursor, I usually move the mouse so that this wand doesn't obscure what I'm writing. It might be a bad habit, but usually programs/browsers didn't mind. I have Opera 36.0.
Well here it is now: http://ist.insecta.pro/?cap=бражник&geof=&geos=
I click after the first letter, move the mouse, and the cursor jumps to the end of the word.

I tried, everything is normal. True, I have the 37th version, but it's unlikely that there are such big differences... And, of course, moving the mouse cursor is quite a normal habit, and sane programs should not object to this.
People, does anyone else see similar problems?

Ah, here I am. It seems that Opera offers to select options in the drop-down list (on its own initiative, not to be confused with auto-substitutions, which are not yet available on the site) and if you hover the mouse over one of the options, the field is highlighted and the cursor runs to the end of the line (such as auto-substitutions in Excel, which sometimes interfere so much). I don't know about disabling this feature (it is clearly from the Opera House invention), but if the cursor is not moved to this selection, but somewhere to the right/up, then everything is bu norm.

Removed the restriction for local names. It wasn't used for Latin before.
Separate fields for the genus and species — essno, realizable. Only this often causes inconvenience, rather than vice versa. Currently, it is searched separately by gender, by species, and by gender/species, but there is only one field. We will soon add auto-substitutions (as in the main search for insecta), and then this field will be more convenient than ever.
But the restriction on taxa (they say, to search only among insects, for example) is yes, it can be useful for the future.

I'm afraid this is the opinion of a person far from zoology.

Write specific examples of search problems in one line.

Ah, here, I caught it. ...

Yes, it seems that's exactly what happened. At least in Chrome, there is nothing similar, and Opera is really trying to tell you something. Well, if you know what the problem is, then there is no problem with the search.

There was a glitch when not all photo previews were displayed from the database. Now
it is treated. P. S. The system now works only on insects. Now there are ~600.000 photos for 56.000 taxa. But the index is a bit old (from the beginning of this year).
Current link: https://insecta.pro/photos" target="_blank">https://insecta.pro/photos
Discussion: https://insecta.pro/community/65657" target="_blank">https://insecta.pro/community/65657

I tried to use your search engine after a long break, and the result, to be honest, was disappointing. I entered "Latibulus argiolus", the result is only five photos. Therefore, I chose the "pupa" stage, the result is zero. At the same time, Google photos give out an order of magnitude more photos of this rider, and a number of the first thirty do not contain "left" inclusions, and for the request "Latibulus argiolus pupa" the picture of the pupa you are looking for is in the first position. And this picture is quite old, I saw it last year, and the year before that. I remember that the results of last year's comparison with Google were more optimistic. Whether Google has caught up, or your search engine has deteriorated... It seems to me that the only situation in which your search engine in its current state will be in demand is the search for taxa with names that resemble female names, because in such cases it is better not to even try to contact Google or Yandex photos.

It's good that there are no sunflowers now, I approve of that.

INO, by the way, you gave a good example to compare the principles of operation of two systems, their advantages and disadvantages.

1. Google eats up a sea of images. It has a lot more pictures of insects than on Insecta.

2. Google is able to analyze the environment of images, including links, plus the images themselves and their similarity to others. As a result, it can respond twice as accurately to your request. And maybe very much lie.
Insecta can do almost nothing compared to Google. But it knows much better where the caption with the taxon is attached to this particular photo, where with the geography, etc.

3. Google doesn't bother much with Latin, taxa, and insects, and uses more or less general algorithms for them. And it pays them a little less attention to the accuracy of selection than to more popular and commercial queries (this point is just my value judgment, if anyone has other observations, let's discuss it).

As a result, the image output page for the Latibulus argiolus query (by the way, these pages are different, but I don't think it's radical) will have something similar at the top of the output, and slag will gradually accumulate at the bottom.

Consider 2 situations:
A) An expert is looking for photos (for example, to illustrate your work or for some other educational purposes). If the preview image shows this at all, then it can quickly separate the slag from the non-slag. If you can't see it, then follow the link from Google to the source for each photo.
B) Photos are not searched for by an expert, but by a doubter. None of the photos gives 100% confidence that Google correctly understood what the person who uploaded/signed the photo on the source site meant. Therefore, you will need to bypass these very sources and only then you can be sure that yes, this is it (more precisely, at least be sure that the person who signed the photo thought so).
Examples are also visible in your sample, and there will be even more of them for relatively rare species (in the form of photos, not in nature, I mean) that have more popular relatives in the same genus.

This is for simple Latin queries.

If we add the stage of development and other geography, the situation with accuracy worsens.

4. Google doesn't know any taxon trees. If you search for a generic name, it can give good results simply because it is part of the full specific name. But it won't be the same for the tribes and other families.
Insecta knows about trees. Therefore, it searches precisely for child taxa (as well as those defined before the requested tribe / family or other taxon level). But. Her knowledge is only as good as the taxon tree on the site. And it is, to put it mildly, not good everywhere (ow, curators! we are waiting for you :--)

5. Google doesn't know synonyms and local names, but it can guess them based on their similarity. Therefore, when requesting Latin, you can see photos signed with just English names, and often this is quite an adequate result. But this is precisely the result of guesswork.
Insecta doesn't know how to guess, but it does have a list of synonyms and local names from the same tree. Which it knows how to substitute, but, again, only within the adequacy of this list. The more accurately users add information, the more accurate the search.

Plus, with regard to the situation right now, Google is constantly updating/adding info, and the index still hangs on the Insectfrom the beginning of the year (and in some places from the end of last year). Of course, the working version should be updated / updated regularly.

And, of course, the image-eating system itself is different. What Google ate in the first place (and did the right thing, taking into account its approach) may end up on Insecta in 5 years or not at all (and this will often also make sense, although sometimes there is simply a lack of resources).

The goal of the search engine on Insecta is to search for images with accurate taxon identification (as accurate as it is in the source), with accurate selection by parameters (also depending on the source), and taking into account the taxon tree (selecting child taxa), as well as synonyms/generic combinations and local names (depending on the adequacy of the tree on the Insecta).
I think that now it all works on the example that we have. And that the images are still not enough compared to Google-yes :--)

As with other similar projects — we often don't have enough feedback , so thank you for your comments!

P.S. Regarding women's names-drop a couple of examples, I think not everything is so bad for Google should be :--)

P. S. S. If you plant a "correct" tree for a certain taxon (for example, a subfamily) on an Insect, the search accuracy increases significantly. And the very principle of operation, its differences from Google, do not go away.

This post was edited by Asar - 01.09.2017 01: 52

P.S. Regarding women's names-drop a couple of examples, I think not everything is so bad for Google should be :--)

That's right, I also saw this. And not only in the case of women's names, but also various products, companies, etc.
Type in Google, say "Veronica" and on what page will you get a photo of the plant?
Or say "Formica" - a picture of the ant will certainly be, but somewhere below.
But in general, this is easily treated by adding the words plant or insect to the query, respectively.

With one-word ones, of course, there will be problems with any intersections from the real world in noticeable quantities. For childbirth, by the way, sp should help well.