I use the platform in academic projects, and my students love it. The platform is powerful, fast, and easy to use. It fully meets the demands required for large-scale projects! The tool is very interesting, especially the feature that allows downloading annotations in various formats such as COCO and YOLO. Overall, I’m very satisfied with the platform. Review collected by and hosted on G2.com.

It serves its purpose very well. I believe the platform could consider supporting less common types of annotations, such as a system for time series annotations. This is just a suggestion, not a criticism of the platform. Review collected by and hosted on G2.com.

Very easy-to-use platform, fast annotation, Sam3, augmentation Review collected by and hosted on G2.com.

I need a private project, but the cheapest plan is too expensive. Review collected by and hosted on G2.com.

Roboflow has proven to be an excellent tool for accelerating computer vision development, offering a highly intuitive workflow that makes it easy to review, relabel, and retrain datasets. I particularly value that it consistently provides access to the latest YOLO versions for training, ensuring that our models stay at the cutting edge. The platform excels in object detection performance, making it straightforward to standardize input to output, while also managing model version control with minimal overhead. Beyond the platform itself, the Roboflow community adds tremendous value through its webinars, blogs, and collaborative resources. The staff has also been consistently professional, knowledgeable, and easy to engage with (big shout out to Salo!), further reinforcing the partnership you have with the company. Review collected by and hosted on G2.com.

The only drawback I’ve encountered is the cost associated with self-hosted inference, which makes scaling deployments more expensive than anticipated. While the platform’s capabilities are strong, this cost factor can become a limiting consideration for larger-scale operations. Review collected by and hosted on G2.com.

The flexibility in dealing with the whole process Review collected by and hosted on G2.com.

There are only a few auto-generated bounding boxes right now, and I’d really like the option to label more photos. Review collected by and hosted on G2.com.

As a researcher, the best feature of Roboflow is its data annotation interface. For our academic paper on leukocyte classification, we heavily relied on it to build our dataset.

The platform made the entire annotation process incredibly fast and efficient. What could have been a very tedious task was streamlined. The collaboration features were also a major advantage, allowing us to easily manage contributions from different team members and maintain consistency in our labels. Review collected by and hosted on G2.com.

Honestly, for our specific use case (dataset annotation and management for a research project), the platform performed perfectly. We did not encounter any significant downsides that hindered our workflow. Review collected by and hosted on G2.com.

The end-to-end Roboflow ecosystem is truly something magical.

The way the main annotation engine learns from your labeling efforts and begins recommending annotations is incredibly powerful. These annotations can then be exported via the Roboflow API in all the standard bounding box formats, compatible with a variety of popular models.

What really stands out is how intuitive the platform is , making annotation, export, and even testing on standardized models incredibly easy. The Supervision and Inference libraries, as extensions of the core Roboflow tools, provide a complete, end-to-end solution for building custom computer vision applications.

The Roboflow community is also extremely supportive. From in-depth YouTube tutorials by Piotr to active discussions, you’re never without help or inspiration. On top of that, Roboflow continues to push boundaries with innovations like RF-DETR, showing their commitment to staying ahead of the curve.

Overall, the Roboflow ecosystem and the vibrant community behind it make it a fantastic platform for anyone serious about computer vision Review collected by and hosted on G2.com.

The only real issue I’ve run into is with exporting large datasets.When the dataset is big, you can sometimes get stuck on the export screen with a “Spinning up the servers, Feeding the Squirrels ” message. That’s the one part of the experience that genuinely bugs me. Otherwise, Roboflow is light-years ahead of similar tools like CVAT or Labelbox.

As an engineer, I understand how hard it is to maintain data consistency across a distributed system , it's no small feat. But even something like a message saying “We’re processing your data , please check back in a few minutes” would go a long way. I don’t mind waiting 5 or 10 minutes, but just staring at a loading screen during that time feels a bit frustrating. Review collected by and hosted on G2.com.

Roboflow makes it simple to label, train, and deploy vision models. The interface is clean and fast. You get prebuilt tools for dataset management, augmentations, and model training. The API lets you move models into production without custom pipelines. You spend less time on setup and more time testing models. Review collected by and hosted on G2.com.

Some features are only available in higher plans, but that also makes it easy to scale when your projects grow. Advanced analytics are limited, but this keeps the platform focused and simple to use. Review collected by and hosted on G2.com.

- Rapid customer support

- Quick annotation and training times for proof-of-concept computer vision models

- The Roboflow Universe makes finding image data or models that could fulfil your use case much easier

- Easy to teach my team how to get started, I use as an pathway into computer vision for them

- Clean APIs for edge deployment and integration into more complex applications Review collected by and hosted on G2.com.

- Unable to access via Chrome (it may just be my devices)

- Inability to change email associated to an account (will need to create a new account with the new email) Review collected by and hosted on G2.com.

The ease of use is certainly the best thing about Roboflow. I've already used a lot of other annotation tools, and Roboflow is by far the best one, considering usability and implementation. By analogy, it's like "plug and play", the upload process is easy and fast; the annotations can be done in a very consistent way (dividing the task between the team); the integration with other tools is also very fluid, offering a wide range of export formats, besides the very functional API. My team and I have used Roboflow for annotating biological images, which has contributed to the development of new solutions in this field and even proposed new methods, as evidenced by our publication of two papers using this material. During that time, we used the annotation tool every day for about three months, annotating more than 1000 images, and we found no trouble in doing so. The experience was very different compared with the usage of Label Studio, an open source annotation tool - though the interface is beautiful, there are a lot of minor bugs that take us a long time to fix in order to actually start doing the work. I haven't used the consumer support yet, but I guess it's good. Review collected by and hosted on G2.com.

The main problem I see in Roboflow is that it's impossible to keep my dataset private in the free plan. I know it's important for you to reduce the functionalities in this plan, but, in my opinion, at least one dataset should be possible to keep private (or maybe offering a subscription only with this feature). Review collected by and hosted on G2.com.

I’ve been using Roboflow for my computer vision projects, and I’m seriously impressed. The platform is extremely easy to use, even for someone who isn’t deeply technical. From uploading datasets to training models, the entire workflow feels seamless and intuitive.

What really stands out is how fast it connects with the API—integration takes just a few minutes, and there’s no unnecessary complexity. The documentation is clear, and everything just works out of the box without endless configuration. Review collected by and hosted on G2.com.

I’d love to see more detailed and flexible tools for annotation, which would make labeling even easier and more precise for complex projects. Review collected by and hosted on G2.com.