Custom Vision Service
Welcome to the Custom Vision Service API Forum
Categories
API – Any ideas or feedback pertaining to features or enhancements to Custom Vision Service API.
Documentation – Any ideas or suggestions for the API Reference or Documentation.
Language Support – Submit a request to have a particular language supported.
Samples & SDK Request – Let us know if you would like to see a Code sample or SDK provided.
Release Notes – Recently updated December 19.
-
Export & Models or Download photos trained
Due to single user limitation, would like to export & import models to other account. Or, wish if photo used for training be downloadable at once so to transplant model easily.
13 votes -
Paid (S0) plan with upper API call limits
We can easily reach limits per day of prediction;
Happy to subscribe paid plan (probably from Azure Portal like as other Cognitive Services APIs) with upper limits of prediction and/or other API calls4 votesHello,
This has been added. Please see: https://docs.microsoft.com/en-us/azure/cognitive-services/custom-vision-service/home#march-1-2018.
Thanks,
Luke -
API in java similar to C#
It is really nice to have the API in Java as well. Something that is similar to the API in C#. Using Http client is not so interesting.
1 voteHello,
Please see our Java Quickstarts:
https://docs.microsoft.com/en-us/azure/cognitive-services/custom-vision-service/java-tutorial
https://docs.microsoft.com/en-us/azure/cognitive-services/custom-vision-service/java-tutorial-od
Thanks,
Luke -
Export Model to Open Neural Network Exchange (ONNX) Format
It would be very useful if you could export the model into the Open Neural Network Exchange (ONNX) format. Allowing the run the model locally on non-apple devices using CNTK or other supported frameworks.
2 votesHello,
This has been added: https://docs.microsoft.com/en-us/azure/cognitive-services/custom-vision-service/home#may-7-2018.
Thanks,
Luke -
Next picture button from picture detail screen
I'd like to be able to advance to the next picture in whatever view I clicked on to get into the picture detail screen without closing out each time.
3 votes -
Select picture from picture detail screen
I'd like to be able to select a picture from the picture detail screen so I don't have to go back to the previous screen just to check the box.
2 votes -
Image Details Screen -Move save and close button to the top
At my screen resolution the save and close button keeps being pushed off the page. I now have many tags being displayed for the same picture on the prediction screen image details and I have to scroll way to the bottom to get to this button. If I change a tag I have to go looking for it way down there just to keep my tags I've added.
1 vote -
Delete a picture from picture detail screen
The images thumbnails are small so i click on it to expand the view confirm that I want to delete this picture but then I have to close that view and go back to the thumbnails check the box and click delete. I want to just delete it from where I am at. All this clicking around is a waste of time.
1 vote -
Training - Allow us to crop or set bounds box for each tag
I'd like the option to upload a single picture and tag it into multiple tags but tell the computer where in the picture to focus on what tag. This could help us get object level tagging if we wanted to take the time to train it.
10 votes -
After Training - Filter for only images under probablity threshold
After training allow a filter to only show the red outlined pictures that don't meet the the probability threshold. I want to review them to see if I've included bad examples in the training set that need to be removed. This process is hard when there are only a few on each screen and I'm going through 10 -20 pages of pictures to find those that should possibly be deleted out of the training data set.
1 vote -
offline prediction for .net
The offline prediction is very vital because users don't need to connect with cloud for every prediction while they are using just their own modules
9 votes -
Orientation of images
The orientation of the image matter when using, I think it is better if there is an option to let the system know to change the orientation and train
4 votes -
Provide managed .DLL for offline classification
It would be very helpful to be able to download the compact model as a managed .DLL that has a single .Classify method. This way we could integrate the real time model into any application.
Maybe add GPU option and/or multicore.
9 votes -
Multiple Users to access a single project
It would be nice to see a collaboration feature - Multiple Users to access a single project . This feature exists with some of the API's undr Cognitive services umbrella.
13 votesHello,
Projects attached to Azure Resources have the permissions of those resources and are sharable. Please see: https://docs.microsoft.com/en-us/azure/cognitive-services/custom-vision-service/home#march-1-2018.
Thanks,
Luke -
Multi languages for Tags
I have an application in multi languages interfaces while it uses the same categories and the same tags .. it will be nice if possible to create tags with many languages
2 votes -
Capable of quality assurance scenarios
It would be great if we can train the service to be able to inspect and detect quality defects on manufactured parts.
2 votes -
Allow temporary "burst" of images
Currently there's a rate limit if you send too many images in a short time (what's the exact limit by the way?). I'm working on a project that does not send lots of image - but it sends about 10 or 20 withing 2-3 seconds. I then only get results for a few of them and get "Rate limit is exceeded" for the others. Would it be possible to have a limit on a daily basis instead of "x per second"?
4 votes -
Tags - boolean operations
It would be helpful to have some tools for boolean operations on tags - for example, if I have an "id" tag I'd like to select all non-id images, and perhaps combine them with other subsets. TOPCAT shows how it might go (see e.g. www.star.bris.ac.uk/~mbt/topcat/sun253/SubsetWindow.html).
3 votes -
Use prediction images for training
Prediction images appear in the web GUI, but I can't see a way to make them available from there for new training iterations (learn as you go). Is there a way - or could one be added?
5 votesYou can Tag images the ‘Predictions’ tab by selecting some images and then clicking on ‘Tag’. You can also open an image and apply multiple tags to it. This will add the image to your training data set. If you want to programmatically tag images, you can use the ‘CreateImagesFromPredictions’ feature of the training API.
-
Statistical tools
It would be great to have access to more some more basic stats on the training process - learning curves, errors, ROCs, AUC values - the more the better.
4 votes
- Don't see your idea?