There seems to be a bug somewhere in the Docker ecosystem (it's probably
either Docker Compose, Docker Buildx or BuildKit) that causes `docker
compose build` to ignore base images that are already present in the
system, and instead fetch them from Docker Hub, if there's a custom
Buildx builder configured. There's a bug report here:
<https://github.com/docker/compose/issues/9939>.
This bug means that when the build pipeline builds the `cvat_ci` image,
it's based on the latest release of `cvat/server` from Docker Hub
instead of the version that we just built. Consequently, we run the unit
tests against that release instead of the development version.
Fortunately, we don't actually need to set up a Buildx builder in most
jobs (including the `unit_testing` job), so just don't do that.
Also, use `cvat/server:local` as the base image in `Dockerfile.ci`. This
will prevent a similar bug from reoccurring in the future, since the
`local` tag should never be uploaded to Docker Hub.
### Motivation and context
This is a very simple pull request. The type of the credentials
parameter of `make_client` is currently `Optional[Tuple[int, int]]`, but
it should be `Optional[Tuple[str, str]]` as used by `Client#login`. This
PR makes that change.
### How has this been tested?
The typing does not affect the functionality of the code (just the
warnings I get in an IDE).
Issue: https://github.com/opencv/cvat/issues/1217
Currently there are a few proposals for SSO authentication to bypass the
current user/password login on the UI. By using Django social accounts
it is also possible to use SSO on the API, retrieving the security token
by passing the code from the OAuth2 workflow. This is an example using
Amazon Cognito, but any other social account could also be added.
### Motivation and context
Currently CVAT has no functionality to log in with SSO. Other current
proposals bypass the current Django framework to add SSO in the UI only,
but still use username and password for the API. Using Django social
accounts integrates SSO with the API as well, allowing it to be used as
an alternative to the username and password, but can also be used
together with other SSO frameworks that are UI only.
### How has this been tested?
Unit tests for SSO manager in cvat-core and integration test with
cvat-sdk for /auth/cognito endpoint.
### Checklist
<!-- Go over all the following points, and put an `x` in all the boxes
that apply.
If an item isn't applicable by a reason then ~~explicitly
strikethrough~~ the whole
line. If you don't do that github will show an incorrect process for the
pull request.
If you're unsure about any of these, don't hesitate to ask. We're here
to help! -->
- [x] I submit my changes into the `develop` branch
- [ ] I have added a description of my changes into
[CHANGELOG](https://github.com/cvat-ai/cvat/blob/develop/CHANGELOG.md)
file
- [x] I have updated the [documentation](
https://github.com/cvat-ai/cvat/blob/develop/README.md#documentation)
accordingly
- [x] I have added tests to cover my changes
- [x] I have linked related issues ([read github docs](
https://help.github.com/en/github/managing-your-work-on-github/linking-a-pull-request-to-an-issue#linking-a-pull-request-to-an-issue-using-a-keyword))
- [ ] I have increased versions of npm packages if it is necessary
([cvat-canvas](https://github.com/cvat-ai/cvat/tree/develop/cvat-canvas#versioning),
[cvat-core](https://github.com/cvat-ai/cvat/tree/develop/cvat-core#versioning),
[cvat-data](https://github.com/cvat-ai/cvat/tree/develop/cvat-data#versioning)
and
[cvat-ui](https://github.com/cvat-ai/cvat/tree/develop/cvat-ui#versioning))
### License
- [x] I submit _my code changes_ under the same [MIT License](
https://github.com/cvat-ai/cvat/blob/develop/LICENSE) that covers the
project.
Feel free to contact the maintainers if that's a concern.
Co-authored-by: Melanie Day <mday@EYEDIAGNOSIS.local>
Co-authored-by: Maria Khrustaleva <maria@cvat.ai>
Co-authored-by: Nikita Manovich <nikita@cvat.ai>
This PR adds an option to specify file to job mapping explicitly during
task creation. This option is incompatible with most other job-related
parameters like `sorting_method` and `frame_step`.
- Added a new task creation parameter (`job_file_mapping`) to set a
custom file to job mapping during task creation
Pull Request regarding Issue #2987
PIL.Image conversion from I;16 to L or RGB are unsuccessful as for now.
See the corresponding Issue in the Pillow GitHub (Opened 2018, so no
changes to be expected)
https://github.com/python-pillow/Pillow/issues/3011
The proposed changes at least fix this issue for the mode 'I;16' and
delivers a possible solution for other modes (eg. I;16B/L/N).
This results in a correct calculation of the preview thumbnail and the
actual image, the annotation will be performed on.
We have used this solution on our own dataset and created annotations
accordingly.
Constantly increasing RAM consumption in cvat_worker_low docker
container while using automatic annotation with detector model in
nuclio.
### How has this been tested?
<!-- Please describe in detail how you tested your changes.
Include details of your testing environment, and the tests you ran to
see how your change affects other areas of the code, etc. -->
It was tested manually
use the OpenCV contour function to extract contour instead of
scikit-image function. to solve the problem of poor drawing of the
contour see the issue 4660
https://github.com/openvinotoolkit/cvat/issues/4660
This is pretty much the same fix applied to f-BRS in #5384. We've been
using HRNet for a while and now an then we receive "500 errors" just as
reported in #5299 when someone forgets to drop the alpha-channel from
our images.
### Motivation and context
The RuntimeError is a little bit different, but comes from the same
issue: RGBA instead of RGB images:
> RuntimeError: Given groups=1, weight of size [16, 3, 1, 1], expected
input[*, 4, *, *] to have 3 channels, but got 4 channels instead.
### How has this been tested?
I created a task with images with and w/o alpha channel, and the
interactor works on both now.
Django REST Framework ignores the Content-Type on request body parts, so
it doesn't know that they are JSON-encoded. Instead, it just tries to
decode each part as if it was an `str()`-encoded value.
Change the encoding to match the decoding. The only type this matters
for is `str`, because `json.dumps` and `str` produce different encodings
for `str` values.
Remove `none_type` from the list of encodable types since, to my
knowledge, there's no way to encode a `None` value as a
`multipart/form-data` part in a way that DRF will understand.
PR removes previews downloading from CS when task creating (skipped in
PR #5478). In addition, I had to change the test to check for the file
name existing in the message when the specified file is not found in the
bucket, because now the preview is no longer downloaded at the stage of
creating a task.
### Motivation and context
Integration of YOLOv7 as a serverless nuclio function that can be used
for auto-labeling. YoloV7 is the SOTA at the time of this PR therefore
it would make sense to support it in CVAT. The integration is quite
simple into CVAT as docker based on Ultralytics YoloV5 with coco
pretrained model (https://github.com/WongKinYiu/yolov7) and a docker
image (https://hub.docker.com/r/ultralytics/yolov5).
related issue: #5548
### How has this been tested?
Automatic annotation was run using YOLOv7 on a custom dataset.
The serverless function was deployed using
```
nuctl deploy --project-name cvat \
--path serverless/onnx/WongKinYiu/yolov7/nuclio \
--volume `pwd`/serverless/common:/opt/nuclio/common \
--platform local
```
Then using the 'Automatic annotation' action the function was tested and
the auto-generated labels were controlled to check that no coordinates
misfit is happening.
### Use custom model:
1. Export your model with NMS for image resolution of 640x640
(preferable).
2. Copy your custom model yolov7-custom.onnx to /serverless/common
3. Modify function.yaml file according to your labels.
4. Modify model_handler.py as follow:
```
self.model_path = "yolov7-custom.onnx"
```
Co-authored-by: Nikita Manovich <nikita@cvat.ai>
Co-authored-by: yasakova-anastasia <yasakova_anastasiya@mail.ru>
In my understanding of https://github.com/nuclio/nuclio/issues/1821, the
Nuctl (1.8.14) CLI is looking for a path that is only valid on a Linux
environment, which it does not find when running via Git Bash (even when
using the Windows version of Nuctl). However, installing CVAT onto a
Linux VM allows Nuctl to locate this path and operate normally. I
initially found this when setting up CVAT myself on Git Bash as per the
given instructions for Windows 10.
(I am still learning how to use GitHub as far as pull requests / forks /
etc work, sorry if this is not the right way to approach this change.
Please let me know if I've missed something important.)
### How has this been tested?
This is only a change to instructions, but I did test this on multiple
machines . As long as the machine is capable of running a Linux kernel
it shouldn't run into any issues.
Added separate workflow that search cache, it will allow us to remove
duplication of this logic in `main.yml`, `full.yml` and `schedule.yml`
! This solution won't work if PR changes search-cache.yml, but this
workflow is supposed to almost never change
This will let users to run their PyTorch code without network access,
provided that they have already cached the data.
### How has this been tested?
<!-- Please describe in detail how you tested your changes.
Include details of your testing environment, and the tests you ran to
see how your change affects other areas of the code, etc. -->
Unit tests.
Turns out that #5456 had a nasty side effect. Session hooks are called
when pytest is run with `--collect-only` (even though no tests are
actually run in this case), and Visual Studio Code periodically runs
`pytest --collect-only` in order to learn what tests exist in the
project. As a result, it keeps restarting the services and restoring the
database in the background.
Work around this by skipping all logic in the hooks if `--collect-only`
is in the options.
I was supposed to do this in #5524 when I was removing them everywhere
else, but I didn't think to search the documentation. Compose override
files must have the same format as the base file, otherwise Docker
Compose rejects them.
Fixes#5544.
Extracted from https://github.com/opencv/cvat/pull/5083
- Added a default arg for task data uploading
- Added an option to wait for the data processing in task data uploading
- Moved data splitting by requests for TUS closer to the point of use
This PR fixes a problem with local running of tests introduced in #5498.
When running tests locally, there can be multiple tags available, while
we need the `dev` tag, which should be created right before tests are
executed.
- Added the missing image tag in the docker run command
Roman Donchenko
Kirill Lakhov
Maxim Zhiltsov
Padraig X. Lamont
Kirill Sizov
ddx-day
Boris Sekachev
vainkop
Mariia Acoca
Maria Khrustaleva
Andrey Zhavoronkov
Kazuki Matsuda
Michael Fujarski
PMazarovich
Waqar Shahid Qureshi
Rodrigo Berriel
Michael Selasi Dzamesi
Hardik Dava
Nikita Manovich
Irrom
Snyk bot
