beat issueshttps://gitlab.idiap.ch/groups/beat/-/issues2016-02-10T17:14:37Zhttps://gitlab.idiap.ch/beat/beat.web/-/issues/389[experiments] Configurator is messing up dataset entries2016-02-10T17:14:37ZAndré Anjos[experiments] Configurator is messing up dataset entriesIf the user selects a dataset in advanced mode, then the correct dataset is not saved.If the user selects a dataset in advanced mode, then the correct dataset is not saved.Final DocumentationAndré AnjosAndré Anjoshttps://gitlab.idiap.ch/beat/beat.web/-/issues/408[experiments] DB protocol sub-sets not assigned correctly in simple mode2016-03-01T17:44:29ZSébastien MARCEL[experiments] DB protocol sub-sets not assigned correctly in simple modeconfiguring a new expe from the
https://www.beat-eu.org/platform/toolchains/smarcel/vein-recognition-example1/1/
on the DB put vein by default we select in simple mode say "Wrist R" then it can be observed in advanced mode
that the...configuring a new expe from the
https://www.beat-eu.org/platform/toolchains/smarcel/vein-recognition-example1/1/
on the DB put vein by default we select in simple mode say "Wrist R" then it can be observed in advanced mode
that the sets are not assigned correctly:
Datasets [Reset]
train
putvein/1/wrist_R/dev_probes
dev_templates
putvein/1/wrist_R/test_templates
dev_probes
putvein/1/wrist_R/dev_templates
test_templates
putvein/1/wrist_R/train
test_probes
putvein/1/wrist_R/test_probes
the same issue was observed with other DBs.
Besides the fact that the user should double-check in advanced mode -- it will be great if the fuzzy matching probably used to check the corespondance here works better.https://gitlab.idiap.ch/beat/beat.web/-/issues/262[experiments] default plotter is not selected and missing plotter parameter2015-08-19T12:51:08ZFlavio TARSETTI[experiments] default plotter is not selected and missing plotter parameterBTAS TutorialFlavio TARSETTIFlavio TARSETTIhttps://gitlab.idiap.ch/beat/beat.web/-/issues/296[experiments] Despite atnt/2 shows on the experiment display, atnt/1 is still...2015-09-08T01:51:19ZAndré Anjos[experiments] Despite atnt/2 shows on the experiment display, atnt/1 is still chosenIf you fork this experiment:
https://www.beat-eu.org/platform/experiments/tutorial/tutorial/full_isv/2/atnt_isv_DCT12x8_100G_U50/
You'll see the datasets for ATNT/2 are chosen for you.
If you run it (the experiment will fail, bu...If you fork this experiment:
https://www.beat-eu.org/platform/experiments/tutorial/tutorial/full_isv/2/atnt_isv_DCT12x8_100G_U50/
You'll see the datasets for ATNT/2 are chosen for you.
If you run it (the experiment will fail, but no worries), you'll see that the toolchain shown on the bottom indicates that ATNT/1 datasets are still chosen in stead of what is displayed. Must be a bug.BTAS TutorialPhilip ABBETPhilip ABBEThttps://gitlab.idiap.ch/beat/beat.web/-/issues/362[experiments] Empty list of datasets when creating a new experiment2015-12-17T11:59:28ZPhilip ABBET[experiments] Empty list of datasets when creating a new experimentSteps to reproduce:
* Create a new experiment
* Click on [Advanced mode]
The dataset comboboxes are all empty.
If you first select a database/protocol combination in "simple mode", the "advanced mode" works correctly.Steps to reproduce:
* Create a new experiment
* Click on [Advanced mode]
The dataset comboboxes are all empty.
If you first select a database/protocol combination in "simple mode", the "advanced mode" works correctly.https://gitlab.idiap.ch/beat/beat.web/-/issues/482[experiments] Environment "previous" relationship is not honoured by experime...2017-12-04T10:47:30ZAndré Anjos[experiments] Environment "previous" relationship is not honoured by experiment configuratorThe "Environment" objects have a "Previous Version" entry that point to environments they are suceeding. This information should be used by the experiment configurator to select reasonable defaults when the user forks an experiment that ...The "Environment" objects have a "Previous Version" entry that point to environments they are suceeding. This information should be used by the experiment configurator to select reasonable defaults when the user forks an experiment that uses outdated environments.
My recollection is that was the case, but if I fork an experiment using "Scientific Python (0.1.0)", the default selector does not point to "Scientific Python (1.0.0)". It would be great if that was the case.
Example: https://www.beat-eu.org/platform/experiments/pkorshunov/pkorshunov/speech-pad-simple/1/speech-pad_lbp_hist_ratios_lr-pa_aligned/
Just fork it and check the default environment on the Global Parameters.Philip ABBETPhilip ABBEThttps://gitlab.idiap.ch/beat/beat.core/-/issues/31Experiment setup() does not handle properly blocks with multiple outputs2017-08-06T11:17:03ZLaurent EL SHAFEYExperiment setup() does not handle properly blocks with multiple outputsFollowing the major revamping of this package, blocks with multiple outputs are not handled in the right way.
In particular, the current implementation of the setup() method for Experiment assumes that a block has a single output (e.g. ...Following the major revamping of this package, blocks with multiple outputs are not handled in the right way.
In particular, the current implementation of the setup() method for Experiment assumes that a block has a single output (e.g. see https://gitlab.idiap.ch/biometric/beat.core/blob/v0.8.1/beat/core/experiment.py#L858).
I am unsure whether there are other implications in the code or not.Tutorial for the FG2015 ConferenceLaurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.web/-/issues/72Experiment setup page error - buttons are greyed out if reset database2014-11-11T08:33:39ZAndré AnjosExperiment setup page error - buttons are greyed out if reset databaseApparently, there is a small error in the JS of experiment configurations. For a forked experiment, if the user resets the database and then sets it to one of the possible values, the "Go" and "Queue" buttons are not reset properly and r...Apparently, there is a small error in the JS of experiment configurations. For a forked experiment, if the user resets the database and then sets it to one of the possible values, the "Go" and "Queue" buttons are not reset properly and remained greyed out.Biometrics Center Kickoff Meeting and BEAT ReviewPhilip ABBETPhilip ABBEThttps://gitlab.idiap.ch/beat/beat.web/-/issues/291[experiments] Experiment was shared, together with toolchain, but configurato...2015-09-08T01:51:44ZAndré Anjos[experiments] Experiment was shared, together with toolchain, but configurator refuses to use itFor example, this experiment was shared with the team system/administrators:
https://www.beat-eu.org/platform/experiments/setup/chichan/chichan/full_mlbphs/1/MLBPH_comp7/
As you can see, if you try to fill-in the label and then pre...For example, this experiment was shared with the team system/administrators:
https://www.beat-eu.org/platform/experiments/setup/chichan/chichan/full_mlbphs/1/MLBPH_comp7/
As you can see, if you try to fill-in the label and then press the Go button, the configurator refuses to submit the experiment saying the toolchain is not public.
The expected behaviour is that it allows us to run the experiment, given the toolchain is also shared with us:
https://www.beat-eu.org/platform/admin/toolchains/toolchain/213/?_changelist_filters=q%3Dchichan
BTAS TutorialSamuel GAISTSamuel GAISThttps://gitlab.idiap.ch/beat/beat.web/-/issues/348[experiments] Green toolchain filling while running2015-12-11T10:30:57ZSébastien MARCEL[experiments] Green toolchain filling while runningWhen running an experiment we don't see anymore the process in "green" on the graphical representation of the toolchain.
This would have to be hooked in into the new setup page.When running an experiment we don't see anymore the process in "green" on the graphical representation of the toolchain.
This would have to be hooked in into the new setup page.Final Documentationhttps://gitlab.idiap.ch/beat/beat.web/-/issues/260[experiments] Hand-picked environment on experiment block does not appear on ...2015-09-03T12:16:47ZAndré Anjos[experiments] Hand-picked environment on experiment block does not appear on experiment displayAt the experiment display page (after the experiment has finished), if the user has chosen to override the environment on a specific block (or the queue for that matter), this information is not displayed on the mid-column with the per-b...At the experiment display page (after the experiment has finished), if the user has chosen to override the environment on a specific block (or the queue for that matter), this information is not displayed on the mid-column with the per-block overrides. That would be a handy feature to have.BTAS TutorialAndré AnjosAndré Anjoshttps://gitlab.idiap.ch/beat/beat.web/-/issues/298[experiments] I can now delete experiments with attestations!2015-09-08T10:33:59ZAndré Anjos[experiments] I can now delete experiments with attestations!This is a critical one!
https://www.beat-eu.org/platform/experiments/tutorial/tutorial/eigenface_with_preprocessing/1/eigenface-with-preproc-15/
@samuel.gaist, can you please have a look? Thanks, AThis is a critical one!
https://www.beat-eu.org/platform/experiments/tutorial/tutorial/eigenface_with_preprocessing/1/eigenface-with-preproc-15/
@samuel.gaist, can you please have a look? Thanks, ABTAS TutorialSamuel GAISTSamuel GAISThttps://gitlab.idiap.ch/beat/beat.web/-/issues/412[experiments] In the editor, boolean parameters aren't correctly displayed2016-03-14T16:34:17ZPhilip ABBET[experiments] In the editor, boolean parameters aren't correctly displayedOn Chrome on Windows, the checkbox of the boolean parameters is very big (see screenshot). In Firefox on Windows, it looks ok.
![Capture](/uploads/367f07367a4fe13f5c0eab8f6b26cc07/Capture.PNG)
Additionally, on both browsers, the do...On Chrome on Windows, the checkbox of the boolean parameters is very big (see screenshot). In Firefox on Windows, it looks ok.
![Capture](/uploads/367f07367a4fe13f5c0eab8f6b26cc07/Capture.PNG)
Additionally, on both browsers, the documentation string of the boolean parameter isn't displayed. I don't have access to my Mac right now to test if its displayed there.https://gitlab.idiap.ch/beat/beat.web/-/issues/254[experiments] Label auto-completion seems broken2015-08-17T21:52:39ZAndré Anjos[experiments] Label auto-completion seems brokenFor some reason, label auto-completion seems broken at the staging platform.
If you try to create an experiment and click on its name, it will auto-complete the full experiment name instead just the final bit:
![image](https://gitl...For some reason, label auto-completion seems broken at the staging platform.
If you try to create an experiment and click on its name, it will auto-complete the full experiment name instead just the final bit:
![image](https://gitlab.idiap.ch/biometric/beat.web/uploads/db7dcd300231202e7167faf5e2663a36/image.png)
I guess this change is related to the recent ``label`` -> ``name`` clean-ups performed.BTAS TutorialSamuel GAISTSamuel GAISThttps://gitlab.idiap.ch/beat/beat.web/-/issues/461[experiments] Little misleading web interface2017-08-06T16:44:51ZTeodors EGLITIS[experiments] Little misleading web interfaceI find the web interface a little misleading when you set up an experiment.
After one chooses a block, 4 buttons appear that enables you to modify parameters, view algorithm, etc. The black button means "Modify algorithm parameters speci...I find the web interface a little misleading when you set up an experiment.
After one chooses a block, 4 buttons appear that enables you to modify parameters, view algorithm, etc. The black button means "Modify algorithm parameters specifically for this block":
![beat_s1](/uploads/4f90b6d3b0a2328d9067aa12fc0b4844/beat_s1.png)
After you "open" the menu, icon changes, intuitively it feels that pressing the button would just close the algorithm parameter selection menu (for 2 months I assumed this), but in reality this means "reset algorithm on this block to their default values":
![beat_s2](/uploads/b7fe95af199721920534e8dd2b3e35cf/beat_s2.png)
IMHO it would be convenient that the user can close the "algorithm parameter selection menu" without parameter reset (imagine that you are working with a large size toolchain with 10-15 blocks, and you want to change the slot count for each of the blocks - if you must keep open all the menus, the already complicated algorithm setup page becomes non-transparent).
Also, an additional fifth button for reset should be introduced for each block.https://gitlab.idiap.ch/beat/beat.web/-/issues/257[experiments] Long name for experiment causes an error2015-08-25T12:36:53ZAndré Anjos[experiments] Long name for experiment causes an errorIf a name with more than 100 characters is chosen for an experiment (using the experiment configuration), the platform shows a pop-up window with some sort of Python exception that is not digestable by the end-user:
![image](https://g...If a name with more than 100 characters is chosen for an experiment (using the experiment configuration), the platform shows a pop-up window with some sort of Python exception that is not digestable by the end-user:
![image](https://gitlab.idiap.ch/biometric/beat.web/uploads/3a4b7850786d378b0e34de66de5069db/image.png)
Ideally, it should point out the problem and let the user choose another name.
This is probably related to the latest ``label`` -> ``name`` changes on experiments.BTAS TutorialSamuel GAISTSamuel GAISThttps://gitlab.idiap.ch/beat/beat.web/-/issues/285[experiments] Multiple database versions are shown at the configurator2015-09-04T14:56:11ZAndré Anjos[experiments] Multiple database versions are shown at the configuratorWe have just inserted a new database version for AT&T, called ``atnt/2``. The insertion works fine and the database listing flawlessly.
Meanwhile, when configuring experiments, both database versions are still display whereas only the...We have just inserted a new database version for AT&T, called ``atnt/2``. The insertion works fine and the database listing flawlessly.
Meanwhile, when configuring experiments, both database versions are still display whereas only the latest one should be.
Ideally, we need this fixed for the tutorial. @philip.abbet: Do you think you can have a look at this ASAP? (thanks!)BTAS TutorialPhilip ABBETPhilip ABBEThttps://gitlab.idiap.ch/beat/beat.web/-/issues/374[experiments] no capture error log ?2016-01-12T14:55:28ZSébastien MARCEL[experiments] no capture error log ?I forked and re-ran the experiments
http://beatweb-staging/experiments/anjos/ivana7c/simple-antispoofing-updated/1/ivanas-experiment-xxx/
as
http://beatweb-staging/experiments/smarcel/ivana7c/simple-antispoofing-updated/1/ivanas-e...I forked and re-ran the experiments
http://beatweb-staging/experiments/anjos/ivana7c/simple-antispoofing-updated/1/ivanas-experiment-xxx/
as
http://beatweb-staging/experiments/smarcel/ivana7c/simple-antispoofing-updated/1/ivanas-experiment-yyy/
sadly it stopped but for the block that failed the captured error log is empty so so way to debug.
Final Documentationhttps://gitlab.idiap.ch/beat/beat.web/-/issues/483[experiments] Not allowed to share unfinished experiments2017-12-04T10:46:41ZPavel KORSHUNOV[experiments] Not allowed to share unfinished experimentsI forked the experiment, modified an algorithm, started the experiment, and tried to share it with @sebastien.marcel but got the error that "This experiment is not done yet", then it went to "page not found" page.
Here is the experiment...I forked the experiment, modified an algorithm, started the experiment, and tried to share it with @sebastien.marcel but got the error that "This experiment is not done yet", then it went to "page not found" page.
Here is the experiment:
https://www.beat-eu.org/platform/experiments/pkorshunov/pkorshunov/speech-pad-simple/1/speech-pad_lbp_hist_ratios_lr-pa_aligned-rr4/
Here is the error, when tried to share it:
![Screen_Shot_2017-12-04_at_11.11.13](/uploads/8e927c930758d1f1805658c26c8abf36/Screen_Shot_2017-12-04_at_11.11.13.png)https://gitlab.idiap.ch/beat/beat.web/-/issues/108Experiments panel at user micro-site is incomplete2015-04-14T09:32:00ZAndré AnjosExperiments panel at user micro-site is incompleteAfter the (very useful) mod to all visible objects displayed, there is a small issue concerning the experiment list: we cannot see who's the experiment owner. It would be good to add this field.
As a matter of fact, it would be better...After the (very useful) mod to all visible objects displayed, there is a small issue concerning the experiment list: we cannot see who's the experiment owner. It would be good to add this field.
As a matter of fact, it would be better to merge "Toolchain" and "Label" into a single heading (called "Label"), for which we print the whole experiment label (i.e. as returned by experiment.fullname()). That would be more consistent with the other panels.Tutorial for the FG2015 ConferenceSamuel GAISTSamuel GAIST