Ango Hub Reviews

Ango Hub has features that make it suitable for image labeling. The “Smart Scissors'' tool especially works quite well for inventory CCTV footage which is what we work with, it speeds up labeling quite significantly. The FrameCut feature also works but it’s not quite as good as the scissors. It also supports multi-layer TIF, which along with JPG is what we use most.

It’s not immediately obvious how to activate or use the AI Assistance features without reading the documentation, as well as other features. Some things about the UI could be improved.

Our datasets are composed of CCTV imaging, and Ango Hub deals with it well, with its handy AI assistance features and support for multi-layer TIF. Hub is affordable enough and it works well with our small team of labelers.

When we found that a small feature would help our workflow, we emailed Ango but thought nothing of it. The next day, the feature we wanted was added. The team’s quick response has been nothing short of amazing and we’re not sure this could have happened with other bigger platforms. Ango Hub has great performance and we’ve been able to complete our text classification tasks quickly and efficiently.

Right now it’s not possible to label on platforms that aren’t desktop. While this is not a huge deal for us, I thought others might want to know. The export formats are currently limited to two, so you might need to do some conversion depending on your situation.

Our model needs to be trained on classifying user reviews, and so far we have no complaints. The team responds quickly and clearly cares about customers. We're able to perform our labeling quickly and efficiently.

We have different models which need to be trained on different data types, one on audio and one on text transcripts. We like that with Ango Hub, we can just use one single platform to label both. Our labeling team is about 10 people and I like that we can see detailed statistics about everything that’s going on. Pricing is also very reasonable compared to other products like it.

There is no function to zoom into audio waveforms, so if your audio file is really long, you’ll lose precision in the cut. I’m hopeful this can be fixed with an update.

Other platforms only have either audio or text, it’s great to have both in one. This can be especially useful to teams dealing with customer sentiment analysis.

The platform performs well even with heavy video tasks. We did video classification with samples in 4K and had no visible issues. I like that we can import our data into Hub without having to change where it’s located on the server, we just give Hub links to where it is now.

Currently the only video file type accepted is .mp4, so we do have to convert videos to this format which can be time consuming. It’d be great if we could directly upload .avi or .mov files.

We have to classify a large chunk of video files and so far, Ango Hub has worked well for us with no big issues to note. Importing files to Hub is instant since we only give it a link to where the file is on our server, plus it saves us from legal headaches. Video classification works as intended and it does the job.

The team was very well organized during our collaboration and progress was clear on a weekly basis

I enjoyed everything, actually. The project went really well

The dataset for computer vision that we develop for waste collecting companies, in order to give feedback to generators on the quality of recyclable waste sorting

Ango Hub is both a versatile and highly customizable platform. While it does offer precise and high accuracy labeling for most of the commodity machine learning application, it has a flexible architecture to be extended for special cases

The only negative feature is in terms of sharing the data, I believe in the future, there would be a drag-and-drop solution to escalate the initiation of the labeling process.

We worked on labeling youtube videos for the summarization task. Currently, there is no available platform that can provide video summarization labels. Moreover, we achieve a high consistency between the labelers which is a highly important aspect.

Our labeling projects consist of millions of images, and Ango Hub’s performance has been great. Adding files to Hub has been nearly instantaneous and everything feels snappy when loading. Huge kudos to the development team for that. Our annotators had a small learning curve at first but are now used to working on Hub and they too are pleased. The AI assistance features also saved us time thanks to the pre-trained models that are offered.

The interface takes some time to get used to, not everything is immediately obvious. Most labeling platforms are not "standardized" among themselves it seems. Going over the docs once can help clear things out.

Anyone that’s involved with ML in retail knows how crucial the data labeling process is. With Hub’s performance we were able to speed up the process significantly, and our annotators are happy using it too.

We’ve been shopping around for an image labeling software for a while and settled on Ango Hub. We chose it because of its great support for complex image labeling tasks. The interface makes it easy to create large polygons, copy-paste them, modify them, etc. Relations and rotated bounding boxes are a great bonus too, as well as the Smart Scissors tool.

Not everything in the interface is immediately intuitive, it took a little at first to understand what everything did. Importing images is fast but requires you to make a .json.

Our labeling needs are more complex than usual, and we found Ango Hub to be more than enough for all of our needs. It handles everything we throw at it well, quickly and with a great UX.

Our workflow requires us to label a large amount of PDF files, and Ango Hub has been excellent for this. Our PDFs are on S3, and importing them to the platform was easy, we just made a .json with links to our files and they were ready. The labeling itself, while there are shortcuts to learn at first, quickly becomes intuitive and after a while we got really fast. You can run full-text searches on PDF files which is a massive time-save.

Some of the shortcuts can be a little clunky and not immediately obvious at first. You do get used to it but there is a learning curve. It's not immediately obvious how the "AI Assistance" plugins are installed and activated.

We train models to automatically categorize and detect entities in documents, so we need annotation to do that. Ango Hub works for the purpose, and the Ango team is really responsive, even going as far as to implement small features just for our use cases. Overall we are happy with it and we’d recommend it.

Ango Hub is good with PDFs. You can have a group of people annotate the same PDF, and it’ll calculate their consensus score. You can highlight text or draw areas around other text, and it’ll do OCR for you. You can draw relations between labels intuitively. Overall its PDF features are ahead, as far as I can see. Other platforms ask you to convert PDFs to images which is just not convenient at all since you lose the text layer.

It’s probably a pet peeve but Ango Hub is not really mobile friendly at the moment. Most users will use the platform from the desktop so it’s not a big issue, but sometimes I need to look at some of my colleagues’ annotations from my phone and it’s not perfect. I’ve emailed them about this and they did say it was in the pipeline though.

We train our model on labeled PDFs, and until now, we couldn’t find a platform that did PDF labeling well. The best way I can describe Ango Hub is “feature-rich.” You can annotate your data in pretty much any way you can imagine.