I'm using CVAT for the annotation purpose. It's very user friendly compare than other tools. I like Bounding box, polygon, segmentation and keypoints parts with CVAT most. It's takes very little time and convenient to use. Also price is cheap compare than other tools. Also it's have ai feature to run auto annotation and it works perfectly. It's very good performance. I can add my labelers to this platform very easily and their support team is very helpful. Yes, I can export and import data with other tools with CVAT perfectly. Usually, my client give me the COCO or YOLO annotations format after that I import it on CVAT. And it works smoothly. Review collected by and hosted on G2.com.

It's quite slow when load too many objects in one frame Review collected by and hosted on G2.com.

1) The continuous upgrades keep improving the tool and make my annotation work faster and higher quality. 2) The filtering options and mass-apply actions make it easy to apply changes across the whole dataset. 3) I also really like the stats views—they help me keep track of class balance. The export formats are standard helping me to fine tune my model easily. Review collected by and hosted on G2.com.

Quite pricey and the UI can be a bit slow at times - especially when running the SAM on a new image. It would be great to have an option of pre-running SAM on the whole dataset to keep it warm. Review collected by and hosted on G2.com.

It has an extensive API that can be used to integrate with other tools. It’s easy to use and affordable. Review collected by and hosted on G2.com.

I’d like to see more role options. For example, a role similar to the manager role, but without the ability to make changes to projects—only to reassign tasks. It would also be very useful if annotators could assign tasks to themselves, so the workflow is more of a pull model rather than a push model.

The filtering feature also feels a bit lackluster. I’d like an easier way to save and reapply commonly used filters. Review collected by and hosted on G2.com.

I find CVAT very easy to use, thanks to its user-friendly UX and UI, which makes getting hands-on with the web application straightforward. When I first started exploring tools for image segmentation and annotations, CVAT stood out as being very user friendly. It not only simplifies the workflow but also helps manage projects effectively with its capabilities to handle multiple tasks and jobs, making the flow of work much easier. The project management features, like assigning and tracking the status of annotations, are particularly useful when working in a team. The tool also proves helpful by facilitating seamless collaboration, as my team could easily pick up and manage their tasks even when I was not available. Additionally, the initial setup of CVAT was straightforward due to the helpful guidelines and approachable UI design, allowing me to understand and carry out my tasks with ease. Review collected by and hosted on G2.com.

I feel like CVAT could improve its automatic annotation capabilities. With other tools like RoboFlow, once I've annotated some frames, the tool can auto-detect and annotate the remaining frames, which I find really efficient, especially for large datasets. Having to manually annotate thousands of images in CVAT is time-consuming. I also think the exporting process needs clarification because I struggled with understanding which export format was right for my task. If there were descriptions provided for each format, it would make it a lot easier to choose the right one for specific applications like segmentation or detection. Review collected by and hosted on G2.com.

Simple to learn, simple to plan and implement. I needed it for a computer vision project — hand-annotating thousands of dental X-rays with polygon masks across multiple classes — and CVAT's annotation tools made that workflow manageable even as a new developer to the space. Strong value for money. Given the annotation and export flexibility — multi-format support, reliable segmentation tools, and responsive support when I've needed it — the subscription cost is easily justified for a solo developer building a real dataset. I haven't found a comparable tool at this price point that covers the same ground. Review collected by and hosted on G2.com.

The main gap is around batch preprocessing. CVAT doesn't have a built-in de-duplication feature, and there's no way to resequence a batch of X-ray filenames — long alphanumeric strings — into a clean, ordered sequence like 01.png, 02.png, etc. I ended up scripting this myself in Colab before uploading. It's not technically difficult, but it's extra overhead on every batch, and a built-in tool for de-duplication and file resequencing would be a genuinely useful addition for anyone managing large image datasets. Review collected by and hosted on G2.com.

The biggest win for us is the model-assisted annotation loop. We run YOLO and SAM2 pre-annotation directly through CVAT, so the work is mostly verifying and correcting rather than labeling from scratch. For a small operation producing the computer-vision training data behind our product, that's the difference between a dataset taking weeks versus days. We also rely on the flexibility of handling multiple annotation types in one tool — we annotate ball and equipment bounding boxes, hockey stick parts, and pose keypoints, and the project/task structure keeps dozens of video tasks organized without friction. Being cloud-hosted means there's no annotation infrastructure for us to stand up or maintain, and the export formats drop straight into our YOLO/training pipeline. Review collected by and hosted on G2.com.

The main thing we hit was that the default exports include only the frames that have been explicitly annotated. For single-frame detection that's fine, but it breaks sliding-window, multi-frame temporal models that expect dense, contiguous frames — we ended up building a dense frame-extraction step around the export to work around it. The keypoint workflow also has a slightly steeper learning curve than plain bounding-box work. None of these are dealbreakers, but they're worth planning for. Review collected by and hosted on G2.com.

I use CVAT to extract data in COCO01 for creating my AI. What I love most about CVAT is its simplicity and formatting. I particularly appreciate the ability to easily draw each contour of the image I want to digitize with the 'polygon' tool, as well as the ability to create as many attributes as I want. Initializing CVAT was quite simple and quick for me. Review collected by and hosted on G2.com.

I think the image loading time should be improved and, if possible, automate the polygon drawing on images. On several images, identical shapes exist. Instead of redrawing the same shapes each time, it would be necessary to automate the recognition of these shapes and generate the drawing. Review collected by and hosted on G2.com.

I use CVAT to label data for my CV projects involving detection, pose estimation, and segmentation. I like how it makes labeling easy and provides flexible export options which save time. The way I can organize projects and tasks is very user-friendly. A major advantage is handing over partial work to third parties without disclosing other projects. I appreciate the ability to export to any format or import from already labeled data, which makes CVAT my default choice. The dashboard helps me easily spot pending tasks and the specific frame that needs work, especially with multiple projects underway. CVAT is the most user-friendly tool I've tried, and I value the privacy of my work. The initial setup was a piece of cake. Review collected by and hosted on G2.com.

I wanted to experience SAM2 and 3 for some projects but I get the feeling that the scheme for using those tools are not fully implemented the way they were meant to be. When I try to use SAM 3 for work using the magic wand menu, for instance, even when I pick an area that is visibly separated from the surroundings, the model or tool seems to get stuck no matter how many times I try. This does not always happen but happens frequently enough that I cannot fully rely on it. Review collected by and hosted on G2.com.

I use this powerful annotation tool for SEM microscopy image annotation. I like the extensive support of prelabelling through foundation models or custom models via API. I use quality tools for reviewing, like quality reports and create issues with a review mode. The ui is nice and there are helpful shortcuts. It is worth the money. Performance is fine, but could be improved and there is the cvat documentation available online, it could be slightly improved as well. Review collected by and hosted on G2.com.

No option for annotation around an object of no interest. For example, an object of no interest lies within an inner circle of an object of interest. I would like to exclude that object.

Shortcut for viewing and hiding the annotations during `Review` mode.

Slow loading times when opening projects, tasks, jobs.

No option for exporting issues together with annotations. Review collected by and hosted on G2.com.

What initially attracted me to CVAT.ai was the ability to use the free cloud version to learn the platform and validate my annotation workflow before making any investment. That allowed me to develop my annotation schema, understand the tool's capabilities, and confirm it met the needs of my research. As the scope of my doctoral project grew, I discovered that CVAT also offers Professional Labeling Services, which made it possible to scale the annotation effort while maintaining the quality standards I had established.

The biggest strength of CVAT.ai was the combination of a flexible annotation platform and a collaborative Professional Services team. My research required far more than basic bounding boxes. I developed a custom annotation schema with five object classes and additional attributes for blur, occlusion, and truncation, along with detailed annotation guidelines and objective quality acceptance criteria. The Professional Services team worked through multiple review cycles, incorporated feedback as the annotation guidelines evolved, and consistently applied the specification across the dataset.

The final dataset integrated directly into my YOLO training pipeline and became the foundation of my doctoral research. The ability to move seamlessly from evaluating the platform on my own to partnering with the Professional Services team made CVAT a valuable part of the project from start to finish. Review collected by and hosted on G2.com.

I honestly didn't encounter any significant issues during my project. The platform met my needs from the initial evaluation through the completion of my doctoral research, and my experience with the Professional Labeling Services team was consistently positive. Any questions or feedback I had during the annotation process were addressed through the normal review cycle, making the overall experience smooth and collaborative. Review collected by and hosted on G2.com.