Systems Frontend issueshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues2024-01-09T15:21:18+01:00https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1746fix polling of running $jobs in asset page2024-01-09T15:21:18+01:00Milan Gressfix polling of running $jobs in asset pageThe polling is currently only done for the update of the preview but not for the running $jobs so sometimes the updating doesn't work
https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1745#note_350456The polling is currently only done for the update of the preview but not for the running $jobs so sometimes the updating doesn't work
https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1745#note_350456https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1739moving playhead-handel in the swimlane produces sometimes buggy bounding boxes2022-10-31T13:47:29+01:00Milan Gressmoving playhead-handel in the swimlane produces sometimes buggy bounding boxesThe real bug (wrong BB at current frame) is mostly non-breaking and easily fixed by a quick play/pause
![Kapture_2022-10-29_at_11.57.15](/uploads/9b23dca81e8c162c542182b8910cd773/Kapture_2022-10-29_at_11.57.15.mp4)The real bug (wrong BB at current frame) is mostly non-breaking and easily fixed by a quick play/pause
![Kapture_2022-10-29_at_11.57.15](/uploads/9b23dca81e8c162c542182b8910cd773/Kapture_2022-10-29_at_11.57.15.mp4)Feature: Swimlanehttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1735Timeline Viewer: optional "remove from files"-question if removing video from...2022-10-26T15:07:32+02:00Hansen, ChristianTimeline Viewer: optional "remove from files"-question if removing video from timeline?No extra navigating to /files needed after uploading a video accidentally:
<br /><br />
![Bildschirmfoto_2022-10-26_um_15.01.50](/uploads/58669a0997bffd4df5b9fda0debe8436/Bildschirmfoto_2022-10-26_um_15.01.50.png)No extra navigating to /files needed after uploading a video accidentally:
<br /><br />
![Bildschirmfoto_2022-10-26_um_15.01.50](/uploads/58669a0997bffd4df5b9fda0debe8436/Bildschirmfoto_2022-10-26_um_15.01.50.png)https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1731Use tracking information in front end2024-01-10T16:37:38+01:00Koch, AntonUse tracking information in front endThe tracking information is returned in place of the bounding box info once the tracking job completes.
You can currently distinguish a simple bounding box from a tracking bounding box like this:
```javascript
// Simple bounding box (4...The tracking information is returned in place of the bounding box info once the tracking job completes.
You can currently distinguish a simple bounding box from a tracking bounding box like this:
```javascript
// Simple bounding box (4x spatial, 2x param)
[x, y, w, h, confidence, class] = data
// Tracking bounding box (4x spatial, 1x param)
[x, y, w, h, id] = data
````
You can also explicitly add `type=bounding_box` or `type=tracking`, the former will query the Seldon Core service, and the latter will throw a 404 if no data exists.Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1728Jobs interval keeps polling2024-01-09T15:21:18+01:00Koch, AntonJobs interval keeps pollingJust noticed this yesterday, no info yet on if and how to reproduce.
The interval polling the jobs on the files page did not stop after the last job was complete and kept polling (job array was empty).Just noticed this yesterday, no info yet on if and how to reproduce.
The interval polling the jobs on the files page did not stop after the last job was complete and kept polling (job array was empty).https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1727Feature request: (visually) connect two or more bounding boxes2022-10-25T18:28:53+02:00Rittershaus, DavidFeature request: (visually) connect two or more bounding boxesThis was a question brought up by the students in Berlin when demonstrating the bounding boxes: if it is possible to connect two or more bounding boxes and to show these connections graphically (question was probably inspired by looking ...This was a question brought up by the students in Berlin when demonstrating the bounding boxes: if it is possible to connect two or more bounding boxes and to show these connections graphically (question was probably inspired by looking at Synchronous Objects). They clearly had in mind that those lines (and therefore also the bounding boxes) are still visible when the video is playing.
I find it interesting to think about what you would annotate: the relation between two or more persons, so in the end something like "person 1 is giving a cue to person 2" is modelled on the data level but also illustrated graphically on top of the video in the front end. Could also be of interest in the analytical work of the MACoDE students (maybe).
Maybe a bit beyond what's possible to achieve within the short remaining timeframe.
Center to center makes more sense compared to this quick mock up:
![Bildschirmfoto_2022-10-25_um_18.16.08](/uploads/d8c8a0dc3a6e104d0b42648c1c5ad558/Bildschirmfoto_2022-10-25_um_18.16.08.png)Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1726Feature request: BG substraction and video overlay ("trio generator")2022-10-25T18:14:39+02:00Rittershaus, DavidFeature request: BG substraction and video overlay ("trio generator")This also came up in Berlin, looking at the Motion Bank "trio generator" for Deborah Hay:
![Bildschirmfoto_2022-10-25_um_17.59.15](/uploads/273a147249f594dabe8b87e6a9dcecd0/Bildschirmfoto_2022-10-25_um_17.59.15.png)
https://vimeo.com/...This also came up in Berlin, looking at the Motion Bank "trio generator" for Deborah Hay:
![Bildschirmfoto_2022-10-25_um_17.59.15](/uploads/273a147249f594dabe8b87e6a9dcecd0/Bildschirmfoto_2022-10-25_um_17.59.15.png)
https://vimeo.com/29619105
Or also the 21 solos versions with the white background:
![1_h1EUZC7SDWmHs7PRh1FIRg](/uploads/e6576d405dd32f641d81a59c59a9fe10/1_h1EUZC7SDWmHs7PRh1FIRg.png)
Something similar seems to be of interest, but maybe the use case is too specific.
However, something that could match the DSHS use case with the comparison (accuracy/ voher-nachher) could make use of it and/or provide the better use case. Comparing two videos by watching them simultaneously is not really possible, because you can't keep an eye on the movement in both videos at the same time. Subtracting the background and overlaying them could be of interest here as well (although the alignment has to be solved as well).
Could potentially be also something that renders stick figures from pose estimation data.
In Berlin they would of course be interested in getting a newly rendered video out of it, which is something that we discussed critically in other contexts (we don't want to turn it into a video editing tool etc.)Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1725Feature request: render travel paths based on selected time range and boundin...2022-10-25T17:48:45+02:00Rittershaus, DavidFeature request: render travel paths based on selected time range and bounding boxesI was showing examples of the work with Deborah Hay + Between Us and talked about what ML can do, so the the students in Berlin started to come up with feature wishes.
I turned one of their ideas into a feature request (which I think i...I was showing examples of the work with Deborah Hay + Between Us and talked about what ML can do, so the the students in Berlin started to come up with feature wishes.
I turned one of their ideas into a feature request (which I think is not too far from things we've discussed before), which is based on the tracking.
1) First, you select a time range
2) Second, you select one or several persons based on the bounding boxes
3) A graphic with the travel paths of these people within the selected timerange is created and saved as an image annotation:
![Bildschirmfoto_2022-10-25_um_17.42.52](/uploads/ef0c2fb8cd11f59633f5fa8129f339cb/Bildschirmfoto_2022-10-25_um_17.42.52.png)
I'm aware that it won't bring the same results compared to using the tracking system and also talked about the limitations to the students, but we think that it might be worth to see if the results are good enough to make use of it in comparisons (like with Deborah Hay) and/or as output that is anyway used creatively outside of the annotation tool.Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1723Cache AnnotationElements in AnnotationTrack2022-10-24T12:32:12+02:00Mittermeier, MaximilianCache AnnotationElements in AnnotationTrackMittermeier, MaximilianMittermeier, Maximilianhttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1722Implement FrameGrid as svg pattern in RulerTrack2022-10-24T12:32:33+02:00Mittermeier, MaximilianImplement FrameGrid as svg pattern in RulerTrackMittermeier, MaximilianMittermeier, Maximilianhttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1720Swimlane performance needs to be optimized for mobile2022-10-25T13:23:31+02:00Mittermeier, MaximilianSwimlane performance needs to be optimized for mobileZooming and navigating the Swimlane is fairly low-fps on mobile, especially when zoomed in far enough for the framebased grid to be visible.
Profiling shows the following distribution of spent time for dragging the Swimlane:
- 72% - Ev...Zooming and navigating the Swimlane is fairly low-fps on mobile, especially when zoomed in far enough for the framebased grid to be visible.
Profiling shows the following distribution of spent time for dragging the Swimlane:
- 72% - EventHandler for the drag
- 13% - Painting
- 5,3% - Styles
- 3,5% - Reflow
Time within EventHandler is split between (% of total Time)
- 43% - Updating the Ruler Grid/Ticks
- 28% - Updating the AnnotationElements
Within Updating Ruler Grid (% of total Time)
- 23% - buildRulerGridSvgElement > toFormat() (creating the label-string for Ruler Ticks)
- 4,4% - Node.removeChild
Within Updating AnnotationElements
- 25% - buildAnnotationElements > mostly spent on fromISO() and toDuration()
- 3,5% - arrangeAnnotationElements > mostly spent on fromISO()
![Screenshot_2022-10-24_at_12.10.42](/uploads/61ab1239e12207ed444276e6e32b532f/Screenshot_2022-10-24_at_12.10.42.png)Feature: SwimlaneMittermeier, MaximilianMittermeier, Maximilianhttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1716chips/tags in annotation item do not wrap2023-01-17T11:46:41+01:00Milan Gresschips/tags in annotation item do not wrapCurrently, the start time & duration chips and other tags like `Selected Person` and `Live` do not wrap and create a scroll container. We're probably going to get more tags in the future… So, I suggest these chips/tags should wrap around...Currently, the start time & duration chips and other tags like `Selected Person` and `Live` do not wrap and create a scroll container. We're probably going to get more tags in the future… So, I suggest these chips/tags should wrap around to the next line, so they are always visible.
![Bildschirmfoto_2022-10-20_um_12.22.56](/uploads/ecaf59e3f7903783816941547be26c94/Bildschirmfoto_2022-10-20_um_12.22.56.png)https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1710Add mb-skeleton as background when "waiting for conversion" in annotation item2022-11-15T20:05:41+01:00Hansen, ChristianAdd mb-skeleton as background when "waiting for conversion" in annotation itemTimeline viewer:
<br /><br />
![Bildschirmfoto_2022-10-19_um_14.55.35](/uploads/adb9fa6052d36911f245949a7637924d/Bildschirmfoto_2022-10-19_um_14.55.35.png)Timeline viewer:
<br /><br />
![Bildschirmfoto_2022-10-19_um_14.55.35](/uploads/adb9fa6052d36911f245949a7637924d/Bildschirmfoto_2022-10-19_um_14.55.35.png)https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1709strange process label behavior after uploading an asset in /files2022-10-29T13:07:25+02:00Hansen, Christianstrange process label behavior after uploading an asset in /filesLooks like three labels are displayed at one:
<br /><br />
![Bildschirmfoto_2022-10-19_um_14.51.15](/uploads/79b97f926b53f457891a939d3b2d8622/Bildschirmfoto_2022-10-19_um_14.51.15.png)Looks like three labels are displayed at one:
<br /><br />
![Bildschirmfoto_2022-10-19_um_14.51.15](/uploads/79b97f926b53f457891a939d3b2d8622/Bildschirmfoto_2022-10-19_um_14.51.15.png)https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1702Add loading indicator to bounding boxes2022-10-18T22:37:33+02:00Koch, AntonAdd loading indicator to bounding boxesThe `BoundingBox.vue` component has an `isLoading` property that can be used to show a spinner or some other indicator.The `BoundingBox.vue` component has an `isLoading` property that can be used to show a spinner or some other indicator.Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1693Add "persons" to system2022-10-17T11:10:36+02:00Rittershaus, DavidAdd "persons" to systemAs soon as the bounding boxes come along with tracking, it makes sense to offer the possibility to add a person's name to the box (see
#1673). However, as @anton.koch said the last Zoom meeting, it would make sense to really have perso...As soon as the bounding boxes come along with tracking, it makes sense to offer the possibility to add a person's name to the box (see
#1673). However, as @anton.koch said the last Zoom meeting, it would make sense to really have persons as entities in the system.
Not sure how this should be best implemented but from my understanding it could be similar to how we want to implement tags? So persons would be a more specified variant of a tag? Could thus be the first step towards introducing tagging...?!Feature: Data Viewshttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1692Use swimlane component to add an overview to Timeline Viewer2022-10-17T11:02:07+02:00Rittershaus, DavidUse swimlane component to add an overview to Timeline ViewerWith some adjustments, it might not be too much work to sort of "add the swimlane to the Timeline Viewer". It just has to render the annotations (only those with purpose "linking" to begin with; one lane) and if you click on one, you jum...With some adjustments, it might not be too much work to sort of "add the swimlane to the Timeline Viewer". It just has to render the annotations (only those with purpose "linking" to begin with; one lane) and if you click on one, you jump to the corresponding "session". Of course date/time format needs some adjustments but otherwise a simple overview that also can be used to navigate the timeline would already be amazing.
Mockup:
![Bildschirmfoto_2022-10-17_um_10.54.31](/uploads/e6adfe799c6b4e8ae37c0bb12ee67370/Bildschirmfoto_2022-10-17_um_10.54.31.png)https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1691Add OSC messaging to system?2022-10-17T11:41:39+02:00Rittershaus, DavidAdd OSC messaging to system?@anton.koch, @maximilian.mittermei: this not for the next months but maybe something we might look at during a next CCL or whenever there is some time for such things.
It came to my mind when I was using the swimlane a bit more. Would...@anton.koch, @maximilian.mittermei: this not for the next months but maybe something we might look at during a next CCL or whenever there is some time for such things.
It came to my mind when I was using the swimlane a bit more. Would it be possible to use annotations as "triggers" that emit OSC messages? So we could for example send triggers to other software like Ableton or lightning desks/computers or a creative coding framework?
Also, both in Berlin and Frankfurt, we've been once speculating with the dance students about the possibility of adding annotations as a sort of "preset" or "score". The context was slightly different, but it might be linked, since you maybe want to use the swimlane without a video if you use it to emit triggers for another software. So you actually would only need a timeline with "trigger annotations". The live annotator already let's you add annotations without a video to which they have to refer but doesn't allow you do anything further with the captured set of annotations, and without syncing a video you don't have access to the swimlane...
Might also be interesting for creative use to make use of the ML-created data. So you could for example stream pose (estimation) data to another application...https://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1686Swimlane: Add handlers for zoom bar2022-10-19T14:48:12+02:00Rittershaus, DavidSwimlane: Add handlers for zoom bar@maximilian.mittermei: I would suggest to add the possibility to use the zoom with handlers to the left and right of the navigation bar (like in the previous version of the system):
![Bildschirmfoto_2022-10-17_um_09.42.01](/uploads/fca0...@maximilian.mittermei: I would suggest to add the possibility to use the zoom with handlers to the left and right of the navigation bar (like in the previous version of the system):
![Bildschirmfoto_2022-10-17_um_09.42.01](/uploads/fca018d8d155b977dd563e90636c3881/Bildschirmfoto_2022-10-17_um_09.42.01.png)
![Bildschirmfoto_2022-10-17_um_09.44.01](/uploads/d9b46563ee431157315c2c36df4ef850/Bildschirmfoto_2022-10-17_um_09.44.01.png)Feature: Swimlanehttps://gitlab.rlp.net/motionbank/applications/systems-frontend/-/issues/1685Highlight related bounding box when hovering chip in annotation item2022-10-19T14:38:47+02:00Hansen, ChristianHighlight related bounding box when hovering chip in annotation item![Bildschirmfoto_2022-10-14_um_17.04.34](/uploads/169761b2049fbfff9dc4447f1a128997/Bildschirmfoto_2022-10-14_um_17.04.34.png)![Bildschirmfoto_2022-10-14_um_17.04.34](/uploads/169761b2049fbfff9dc4447f1a128997/Bildschirmfoto_2022-10-14_um_17.04.34.png)Feature: Data Views