Datature Reviews

Pros

Datature offers a range of tools for annotating data and training models that can be deployed in production. The team is incredibly supportive and responsive, which I find invaluable. The annotation software is also simple and instinctive, making it easy to work with.

Cons

There's not much to say. The software is growing steadily, and bugs are being fixed while new features get rolled out daily.

Pros

The team behind the software is both supportive and efficient. Navigating the software's interface is simple, and there are plenty of tools available to enhance my experience and cut down on work time. Exporting and testing models using the provided scripts and documentation is also a seamless process.

Cons

While it's simple to export trainings, you're restricted by the computing minutes. This can lead to wasting valuable time on failed experiments. Still, it's understandable, as this is a tough issue to fully avoid and somewhat expected.

Pros

The interface is super user-friendly, and the customer support is fantastic, responding within hours of my requests. I was able to build a deep learning machine vision model in just a few minutes, which was impressive.

Cons

There's really nothing to complain about. I haven't had any problems with the product or service.

Pros

The software has an interface that makes it simple to train neural networks without having to dive into Python code. Their customer service is exceptional, and I've received dedicated help to set up pipelines for my projects. The documentation and how-to videos are also clear and informative.

Cons

At the moment, there's no support for importing existing neural networks, and the software doesn't handle depth maps yet, but I've heard they're working on making both features available.

Pros

Before using Datature, I had to rely on Seaborn to plot graphs and manually run predictions to see the results. Now, everything is integrated, loss curves, prediction visualizations across checkpoints, and even a built-in confusion matrix. This has been so beneficial for my project, as I can immediately identify areas for improvement in my dataset or model settings.

Cons

It would be helpful if I could compare some of the loss curves between different training sessions.

Pros

Datature offers a variety of export formats tailored to different deployment environments, making it easy to switch between formats without retraining models on different architectures. Their feature set for cloud and edge deployment is also really strong.

Cons

Optimization features like quantization are not part of their default export options, but I've been informed that this can be integrated as a custom feature in their Enterprise plan.

Pros

The entire process, from uploading images to annotating and training models, was so smooth and required very little effort. I managed to set up the full pipeline in days, whereas my previous custom implementation took months. Datature's support for a wide range of CV applications, from classification to object detection, segmentation, and keypoints, has greatly simplified my workflow since I no longer need custom code for each application.

Cons

A software tour for new users would be nice, but since the process is clearly categorized step by step, it was not too hard for me to get familiar with it.

Pros

Datature Nexus allows me to go from idea and data collection to a fully functional model in a short time. I especially like their similarity search feature, which makes it easy to filter out images where people are not wearing masks, so I can focus on annotating the ones that do.

Cons

I would've liked their model recommendation system to consider the dataset aggregation statistics as part of the recommendation process.

Pros

It is extremely easy to use, making it ideal for beginners who don't have much expertise in deep learning. It's packed with features, and its data tools are very efficient, which is perfect when you're working with limited data. I love how practical and smooth it is. Plus, it's great for teamwork since you can collaborate with your team online.

Cons

The Free plan has limited training minutes, so there's not a lot of room for experimentation. However, it's still enough to quickly start building a working model.

Pros

The easy-to-use training pipeline on Datature helped me develop a working computer vision model in just half a day. What used to take me a week, wrestling with coding bugs, is now as simple as connecting blocks and tweaking hyperparameters. I'd definitely recommend it if you're having trouble with Jupyter notebooks.

Cons

Some hyperparameters are not adjustable at the moment, but I was told that full customization is coming soon!

Pros

Datature is my go-to tool for annotating complex objects. It's incredibly easy and produces high-quality segmentation across different datasets and use cases. Its simplicity has made it a staple for my team, and we use it regularly for all our labeling needs. The software also supports a wide range of annotation formats, making it easy for us to transition and onboard our existing data.

Cons

Datature Sometimes is not pixel-perfect when handling tricky objects, but their AI refine tool usually fixes most of the issues.

Pros

The support team did a great job onboarding us to their Nexus software, and it's transformed the way we work as a team. Instead of doing tasks like annotations and experiments individually, we now have a customizable workflow that allows us to delegate tasks and review each other's work to avoid mistakes.

Cons

I'd love to have an option where I can sample a few images for QC review, so I don't have to go through the entire dataset manually.

Pros

We use the Nexus software extensively for PhD research at the National University of Singapore. It's incredibly easy to set up and use, allowing us to train and deploy various models quickly without needing to dive deep into technical details. The software also offers customization options, providing flexibility that other softwares don't. Plus, the pricing for annotation is reasonable.

Cons

The school has several AI research projects happening simultaneously, each with its own complexities. A software that provides end-to-end features and allows researchers to fine-tune the training and deployment process is incredibly powerful. After evaluating multiple options, we chose Datature, its Nexus software, and Intellibrush, which fit our needs perfectly.

Pros

This software offers efficient tools for labeling data, allowing me to easily train models with various customization options. The way model evaluation metrics and checkpoint performances are displayed gives me a clear understanding of how the model is performing.

Cons

Honestly, I don't have any complaints! It works exactly the way I want it to.

Pros

The software offers several tools for quickly labeling data, and users can easily train and customize models. It also provides clear metrics and performance checkpoints that show how well the model is doing.

Cons

Everything's running smoothly; I haven't faced any issues or difficulties.

Pros

Datature's customer support is top-notch. When I had trouble training an object detection model, they arranged a Zoom call to assist me. Training models on the software is simple and easy, with plenty of options for customization that I really appreciated.

Cons

I don't really see any downsides. Maybe the free tier doesn't offer a lot of space for experimenting, but if they could introduce a student license, that would be fantastic.