*this is an initial PR, let me know of any changes that need to be made.

• modified structure of package to be initialized with Roboflow() object
• added version objects to detect versioning
• added retrieval for workspaces
• modified all routes to reflect no workspace prefix in front of project name
• changed tests to reflect changes
• changed Project() object to have list of Version() objects

I am trying to upload annotated dataset on roboflow but it takes a lot of time and becomes unresponsive at the end and all the files are being lost and i have to upload them again from the start. I have tried several times but i am unable to upload the data.

On Colab:

!pip install -q roboflow

.
.
.
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
google-colab 1.0.0 requires requests~=2.23.0, but you have requests 2.27.1 which is incompatible.
datascience 0.10.6 requires folium==0.2.1, but you have folium 0.8.3 which is incompatible.
albumentations 0.1.12 requires imgaug<0.2.7,>=0.2.5, but you have imgaug 0.2.9 which is incompatible.

The model.predict() function returns an error when an image with the jpeg extension is used.

Traceback (most recent call last):
  File "/Users/james/src/python/roboflow-python/app.py", line 17, in <module>
    prediction = model.predict("photo.jpeg")
  File "/Users/james/src/python/roboflow-python/roboflow/models/object_detection.py", line 177, in predict
    _, buffer = cv2.imencode(os.path.splitext(image_path)[-1], image_path)
cv2.error: OpenCV(4.6.0) :-1: error: (-5:Bad argument) in function 'imencode'
> Overload resolution failed:
>  - img is not a numpy array, neither a scalar
>  - Expected Ptr<cv::UMat> for argument 'img'

When a jpg image is used, there is no error and inference runs as expected.

This was produced using the latest version of the roboflow library on macOS.

Description

This PR adds a GitHub action to automatically push a new release on pypi when master is pushed.

Moreover, this PR adds a new branch develop that must be used to push new changes before cluster them into a big PR to main.

Let's discuss this new way to collaborate on this project

Description

Users were previously required to use all conditions for active learning so added edge case handling to the conditionals object in case users don't specify all possible key words.

Type of change

Please delete options that are not relevant.

• [x] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] This change requires a documentation update

How has this change been tested, please provide a testcase or example of how you tested the change?

Ran a series of unit tests with the below code snippet making sure all variations were supported including no config object, one field, all fields, some fields, value == None, etc.

from roboflow import Roboflow

rf = Roboflow(api_key="BLlkFnwfSaRFUXBfU0tJ")

upload_conditions_config = {
    "target_classes": ['helmet'],
    "confidence_interval": [30,90]
}

rf.workspace().active_learning("./test_cases", ".png", ["hard-hat-universe-0dy7t", 3], "merge_tester", upload_conditions_config)

Docs

• [x] Docs updated? What were the changes: Added a note in the active learning section that you don't have to use all fields.

Description

I've noticed that on certain versions of python in my case 3.10.7 kiwisolver build was failing as it required C++ 14.0 or greater build tools, that is unacceptable as this should be a simple install. So I've relaxed the kiwisolver dependencies a bit.

List any dependencies that are required for this change. kiwisolver>=1.3.1 (instead of kiwisolver==1.3.1)

Type of change

Please delete options that are not relevant.

• [ x ] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] This change requires a documentation update

How has this change been tested, please provide a testcase or example of how you tested the change?

Relaxed kiwisolver would help ease install for Windows users. YOUR_ANSWER

Any specific deployment considerations

None

For example, documentation changes, usability, usage/costs, secrets, etc.

Docs

• [ ] Docs updated? What were the changes:

Description

This adds Instance Segmentation functionality to the python pip package. I also did a small bit of refactoring but tried to avoid going wild.

Type of change

• [x] New feature (non-breaking change which adds functionality)

How has this change been tested, please provide a testcase or example of how you tested the change?

A couple unit tests have been added. I also tested the new functionality manually via:

import os
import roboflow

ROBOFLOW_API_KEY = os.environ.get("ROBOFLOW_API_KEY")
WORKSPACE_NAME = os.environ.get("WORKSPACE_NAME", "")
PROJECT_NAME = os.environ.get("PROJECT_NAME")
PROJECT_VERSION = os.environ.get("PROJECT_VERSION", "1")

LOCAL_IMAGE = "<local image path>"
HOSTED_IMAGE = "<remote image url>"

# Instantiate Roboflow object with your API key
rf = roboflow.Roboflow(api_key=ROBOFLOW_API_KEY)

# List all projects for your workspace
workspace = rf.workspace()
print(f'workspace: {workspace}')

# Load a certain project, workspace url is optional
project = rf.project(PROJECT_NAME)
print(f'project: {project}')

# List all versions of a specific project
project.versions()
print(f'versions: {project.versions()}')

print(f'images: {project.images}')

# Upload image to dataset
# project.upload(LOCAL_IMAGE)

# Retrieve the model of a specific project
model = project.version(PROJECT_VERSION).model
print(f'model: {model}')

# # predict on a local image
# prediction = model.predict(LOCAL_IMAGE)
# print(prediction)

# Predict on a hosted image
prediction = model.predict(HOSTED_IMAGE) # , hosted=True)
print(prediction)

# Plot the prediction
prediction.plot()
print(prediction.plot())

# Convert predictions to JSON
prediction.json()
print(prediction.json())

# Save the prediction as an image
prediction.save(output_path='predictions.jpg')

Any specific deployment considerations

Once this is merged, we will need to update the docs that are displayed on instance segmentation projects.

eg:

Infer on Local and Hosted Images

To install dependencies, pip install roboflow

from roboflow import Roboflow

rf = Roboflow(api_key="API_KEY")
project = rf.workspace().project("MODEL_ENDPOINT")
model = project.version(VERSION).model

# infer on a local image
print(model.predict("your_image.jpg").json())

# infer on an image hosted elsewhere
print(model.predict("https://example.com/your_image.jpg").json())

# change the confidence for your prediction
print(model.predict("your_image.jpg", confidence=60).json())

# visualize your prediction
model.predict("your_image.jpg").save("prediction.jpg")

Docs

• [ ] Docs updated? What were the changes:

Description

I noticed that the github workflows were not working in https://github.com/roboflow-ai/roboflow-python/pull/51 There seems to be an issue referencing the branch since it exists on the forked repo rather than the main repo.

Run actions/[email protected]
Syncing repository: roboflow-ai/roboflow-python
Getting Git version info
Temporarily overriding HOME='/home/runner/work/_temp/c9897c44-666d-4cac-b37c-b1cf27451ab3' before making global git config changes
Adding repository directory to the temporary git global config as a safe directory
/usr/bin/git config --global --add safe.directory /home/runner/work/roboflow-python/roboflow-python
Deleting the contents of '/home/runner/work/roboflow-python/roboflow-python'
Initializing the repository
Disabling automatic garbage collection
Setting up auth
Fetching the repository
  /usr/bin/git -c protocol.version=2 fetch --no-tags --prune --progress --no-recurse-submodules --depth=1 origin +refs/heads/accept-JPG*:refs/remotes/origin/accept-JPG* +refs/tags/accept-JPG*:refs/tags/accept-JPG*
  The process '/usr/bin/git' failed with exit code 1
  Waiting [13](https://github.com/roboflow-ai/roboflow-python/actions/runs/3063095031/jobs/4944798073#step:2:14) seconds before trying again
  /usr/bin/git -c protocol.version=2 fetch --no-tags --prune --progress --no-recurse-submodules --depth=1 origin +refs/heads/accept-JPG*:refs/remotes/origin/accept-JPG* +refs/tags/accept-JPG*:refs/tags/accept-JPG*
  The process '/usr/bin/git' failed with exit code 1
  Waiting 17 seconds before trying again
  /usr/bin/git -c protocol.version=2 fetch --no-tags --prune --progress --no-recurse-submodules --depth=1 origin +refs/heads/accept-JPG*:refs/remotes/origin/accept-JPG* +refs/tags/accept-JPG*:refs/tags/accept-JPG*
  Error: The operation was canceled.

After making the change to remove the with: block from the Checkout step, the github action seemed to run properly, however tests failed with the following error:

Run python -m unittest
upload and label your dataset, and get an API KEY here: https://app.roboflow.com/?model=undefined&ref=undefined
E

======================================================================
loading Roboflow workspace...
ERROR: tests.test (unittest.loader._FailedTest)
----------------------------------------------------------------------
ImportError: Failed to import test module: tests.test
Traceback (most recent call last):
  File "/opt/hostedtoolcache/Python/3.9.[13](https://github.com/roboflow-ai/roboflow-python/actions/runs/3063184200/jobs/4944979507#step:6:14)/x64/lib/python3.9/unittest/loader.py", line 436, in _find_test_path
    module = self._get_module_from_name(name)
  File "/opt/hostedtoolcache/Python/3.9.13/x64/lib/python3.9/unittest/loader.py", line 377, in _get_module_from_name
    __import__(name)
  File "/home/runner/work/roboflow-python/roboflow-python/tests/test.py", line 35, in <module>
    class TestQueries(unittest.TestCase):
  File "/home/runner/work/roboflow-python/roboflow-python/tests/test.py", line 38, in TestQueries
    workspace = rf.workspace()
  File "/home/runner/work/roboflow-python/roboflow-python/roboflow/__init__.py", line 88, in workspace
    the_workspace = self.current_workspace
AttributeError: 'Roboflow' object has no attribute 'current_workspace'


----------------------------------------------------------------------
Ran 1 test in 0.000s

FAILED (errors=1)
Error: Process completed with exit code 1.

This seems to be due to to the main repository's secrets not being accessible by my forked code. Changing the pull_request workflow to a pull_request_target workflow, may fix this... but it also may introduce vulnerabilities. A better fix might be to mock API requests rather than make real ones.

Type of change

Please delete options that are not relevant.

• [x] Bug fix (non-breaking change which fixes an issue)

How has this change been tested, please provide a testcase or example of how you tested the change?

we'll see what github actions say 😄

Any specific deployment considerations

No

Description

While testing image uploading, I discovered that the logic to check whether an image is in an acceptable format was rather naive and buggy. This should make it a bit more robust.

I also added some tests to prove the bug and its fix. While doing that, I refactored the test suite a bit to allow for more extensibility... and then I went down a bit of a rabbit hole once i realized that secrets couldn't be read and used from a forked PR. I'm happy to break test refactoring out into its own PR if that makes more sense.

Fixes #50

Type of change

Please delete options that are not relevant.

• [x] Bug fix (non-breaking change which fixes an issue)

How has this change been tested, please provide a testcase or example of how you tested the change?

I wrote and ran tests to prove the bug and then prove the fix. I also manually tested that I could upload an image with the capitalized JPG extension and saw it in the Roboflow web UI.

Any specific deployment considerations

Nope

I have followed the ROBOFLOW-tensorflow-object-detection-faster-rcnn notebook.At line get tensorboard link i get this error.Anyone can say what the problem ?

Description

There was an issue because we changed our data structure recently. The resulted in json_response["image"] not existing.

Used PIL to get image dimensions and reconstructed img_dims.

No dependency changes

Type of change

Please delete options that are not relevant.

• [X] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] This change requires a documentation update

How has this change been tested, please provide a testcase or example of how you tested the change?

Tested on two TRT scripts and with three different models. Have yet to test it on an edge device, but this fix should not be device specific.

Any specific deployment considerations

Reading image for image dimensions takes time, if there is a more efficient way to get img_dims or if we are already pulling dims somewhere else in the script. We should pass the variable instead of creating another variable.

Description

After labelling all the images in the dataset, when generating a new dataset version and downloading it, the data.yaml created additional non-existing classes.

Steps to reproduce

1. Label your data
2. Create a new dataset version
3. Check the Modify Classses pre-processing step to make sure you have the right classes
4. Generate the dataset version
5. Download the dataset
6. Check the classes in data.yaml

Expected result

You should only get the classes that you labeled

Actual result

The roboflow creates additional non-existing classes (this also alters the class_id of the real classes)

Note: While labeling, there were some classes that were created by mistake, but they were deleted afterwards since they were never used. This somehow was still picked up by the roboflow.

Evidence

You can clearly see I had 2 classes:

When downloading the dataset, its data.yaml config file looked like this:

This remapped my original class_id's of course. This was my original data.yaml:

Description

This PR adds CLI auth compatibility

We check for configuration from:

1. env vars
2. CLI config
3. defaults

It is compatible with the old Roboflow(api_key="") and now a new Roboflow() object

You can test functionality with the CLI roboflow auth and then

from roboflow import Roboflow

rf = Roboflow()