Now it tells me
rabbitmqlost | 2019-10-29 09:02:26.326 [info] <0.985.0> connection <0.985.0> (172.20.0.4:34330 -> 172.20.0.3:5672): user 'guest' authenticated and granted access to vhost '/'
rabbitmqlost | 2019-10-29 09:02:26.352 [info] <0.1001.0> accepting AMQP connection <0.1001.0> (172.20.0.5:53722 -> 172.20.0.3:5672)
rabbitmqlost | 2019-10-29 09:02:26.355 [info] <0.1001.0> connection <0.1001.0> (172.20.0.5:53722 -> 172.20.0.3:5672): user 'guest' authenticated and granted access to vhost '/'
rabbitmqlost | 2019-10-29 09:05:08.020 [error] <0.813.0> closing AMQP connection <0.813.0> (172.20.0.4:34306 -> 172.20.0.3:5672):
rabbitmqlost | missed heartbeats from client, timeout: 60s
rabbitmqlost | 2019-10-29 09:05:16.634 [error] <0.889.0> closing AMQP connection <0.889.0> (172.20.0.5:53706 -> 172.20.0.3:5672):
rabbitmqlost | missed heartbeats from client, timeout:
Hi @clone95
Thank you for the info and sorry for the late reply.
Interesting that windows requires a different mounting point.
With a new and fresh installation under windows I could now also cause the error you described at the beginning.
With the change of the mounting folder of /var/lib/rabbitmq ---> /var/lib/rabbitmq/mnesia
it worked for me with windows.
I can't explain that it still doesn't work for you, because I can't cause this error anymore.
However, it might help to delete all docker containers (especially the rabbitmq container).
Alternatively, you can try to reinstall the entire LOST installation (but you can keep the docker container images).
Please keep in mind that in .docker-compose.yml the mounting point is the correction you suggested in the rabbitmq container.
This will be further investigated in the coming days.
It would be nice if you could create an issue on Github.
Thank you very much and best regards,
Gereon
Hello @yc10086,
The message you described means that the container with the database (db-lost) was obviously not started.
In Windows, Docker specifically asks for a file share for each container when, for example, docker-compose up starts all containers.
Maybe you missed one of these messages ? This should appear three times (for each container separately).
If one of the containers doesn't get a share, it can't start. So the whole application cannot be started.
Best regards,
Gereon
Arguments available
quick_setup docker
Hello @yc10086,
unfortunately I have little experience with Docker on Windows.
When I installed Docker with Windows, I remember that Docker explicitly asked me to do this when I started the container.
Here is a guide on how to share your Windows drives with Docker Containers:
https://token2shell.com/howto/docker/sharing-windows-folders-with-containers/
Best regards,
Gereon
Hello @mh-mazen
Integration without a docker becomes very difficult.
Please describe your problem in detail.
Thank you and best regards,
Gereon
@glemarivero
Hello
In SIA, you can scroll back and forth through your annotations as long as the annotation task has not been completed and further processed by the pipeline.
Since version 1.0.0 you can also undo your last annotations in MIA (within a browser session).
However, this feature cannot really be used for a review function.
If you want to review the annotations of e.g. another person, then it is recommended to write a pipeline, which consists of two annotation tasks and whose annotations are then merged in a following script (with own logic).
How to create your own pipelines and scripts is described here:
https://lost.readthedocs.io/en/latest/all_about_pipelines.html
In addition, a review tool is planned for SIA, which will allow you to filter for annotations with specific properties.
Best regards,
Gereon
Hi,
Sometimes we have two different types of labels we want our annotators to annotate. For example, given a face, we would like to label its shape, but also its tone. So I created one label that is called Face that has two children (Tone and Shape) and each of them have n and m labels associated. Is there a way to perform both labeling at the same time? I thought about using multi-label pipeline, but that could potentially allow to have two different types of shapes at the same type for example.
So ideally what I'm looking for is to have two dropdown menus, one for each set of labels.
Finally, when do you plan on adding the review tool that would allow us to filter?
Thanks!
docker-machine start default
I am upgrading from LOST version 0.0.6 to LOST version 1.1.0, how can do that while preserving ~/lost/data folder with previous tasks, etc ?
Currently I'm getting:
Original exception was: (_mysql_exceptions.OperationalError) (1054, "Unknown column 'is_junk' in 'field list'") [SQL:
'INSERT INTO image_anno (anno_task_id, timestamp, timestamp_lock, state, sim_class, frame_n, video_path, img_path, result_id, iteration, user_id, anno_time, is_junk, description) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)'] [parameters:
(570, None, None, 1, 1, None, None, 'data/media/folder/.~lock.images.csv#', 1152, 0, None, None, None, None)] (Background on this error at: http://sqlalche.me/e/e3q8) PipeElementID = 2912
git diff 0.0.6...1.1.0 -- backend/lost/db/model.py. Since we introduced multi-label support in SIA you will also need to adjust your scripts: use git diff 0.0.6...1.1.0 -- backend/lost/pyapi/inout.py to see the changes. Finally you need to adjust your pipeline definition files, especially the SIA config part.
Good idea to make the review step mandatory, what i can think of is to add a 2nd Annotation Task in the pipeline (after the original Annotation Task ) and assign it to yourself, then you can review / re-annotate the images.
Not sure if this is really a good approach, but hope it helps :)
Hi there. Got a question. we are evaluating lost for a labelling solution. we have all the datasets in MinIO and access it via Object storage. Is there ways to access via MinIO or S3 related api's
Hi @satnair I am not into MinIO, but if you can mount your datasets/images into the lost filesystem it should work. In order to do that you need to adjust the docker-compose.yml -> under volumes -> add your mount point: my_mount_point_on_host_system:/home/lost/data/media/my_datasets
1 reply
Hi, is there any support for oauth integration?
HI @shruthidipali oauth integration is currently not supported
Hi one more question.......We want to add a textbox to capture the actual value of the ROI...in addition to the label for ROI...any quick suggestion how to do this
Hi @satnair, Unfortunately, there is no quick fix for this... We need to adjust SIA here ... Could you write a feature request on GitHub
Hey, one of my annotators was wondering if there was a way to extend the time that they stay logged into LOST, anything that would enable that?
Hi @burrbank, in the .env config file you can add SESSION_TIMEOUT=x, where x is the time in minutes, until an inactive annotator get automatically logged out