Getting meaningful coaching feedback in volleyball, or any sport, has always required access to a qualified coach. For the hundreds of millions of recreational and developing players who train without that access, technique errors go uncorrected and improvement stalls. Younalize was founded to change that. Built by developer Fabian van Schevikhoven and veteran volleyball coach Matt van Wezel in the Netherlands, the app uses computer vision and biomechanical analysis to deliver expert-level technique feedback to any player with a smartphone. CVAT Online sits at the heart of the data pipeline that makes it possible.Using computer vision to make sports coaching more accessibleThe idea began with co-founder Matt van Wezel, an experienced volleyball coach who has worked with national programs and teams in several countries. He initially approached co-founder Fabian van Schevikhoven with a marketplace concept: international players spend significant time traveling and waiting in airports, so perhaps they could be paid to review videos and give developing athletes technical feedback. The coaching method made sense. Matt often paused a video at the exact moment a player jumped, contacted the ball, or moved an elbow into the wrong position. A still frame gave the athlete a clear point of reference that was easier to understand than general feedback over a moving clip. The economics were harder. The athletes who most needed accessible coaching were not necessarily able to pay elite players for repeated reviews. That pushed the founders toward computer vision. Early experiments with pose estimation showed that body keypoints could be used to measure joint angles and movement phases. Ball detection added the context needed to understand when contact occurred. The concept evolved from a human review marketplace into an AI-assisted coach that could provide consistent feedback on demand. Bringing AI coaching to the volleyball courtYounalize is built around a four-step workflow. A player opens the app, selects a technique (jump serve, float serve, forearm pass, overhead pass, spike, or one of seven total techniques currently supported), records or uploads a short video clip, and within seconds receives annotated feedback overlaid directly on the footage.The analysis pipeline runs on a GPU server. When a video is uploaded, a pose detection model maps the player's body across every frame, tracking 133 keypoints from head to foot. In parallel, a custom object detection model locates the ball (whether yellow and blue, white, or beach-regulation red) even at the moment of contact, when motion blur is at its worst. The raw keypoint and ball position data is then passed to a rules engine, where a set of biomechanical conditions written by expert coaches determines which feedback to surface and what to say. For the jump serve alone, that rules engine contains more than 250 individual conditions, covering every phase of the movement from approach to follow-through.The result is coaching feedback that is both visual, with frame markers showing exactly where a joint is misaligned, and textual, surfacing a specific, actionable correction. The entire process, from video upload to annotated result, runs in seconds.The challenge of training on real-world sports footageThe obvious approach to AI sports coaching is pattern recognition: train a model on thousands of videos of correct technique and flag when a player deviates from the norm. Younalize evaluated this approach and ruled it out early, for a fundamental reason. The app's core users are recreational and developing players: athletes who are performing most techniques incorrectly, which is precisely why they are using the app. A pattern-matching model trained on professional footage has no stable reference point for a player who barely jumps, keeps their arm bent at contact, or has never learned to plant their feet before a serve.The team also tried training sequence-detection models to identify technique phases automatically. The variability across skill levels was too high. A dataset spanning elite club players and first-time beginners produced too much noise for any single model to segment reliably.Younalize's solution separates perception from coaching logic entirely. The AI handles what it does reliably: detecting and tracking bodies and objects in video. The coaching rules live in a separate layer, written by expert coaches and grounded in biomechanics rather than pattern matching. This architecture gives Younalize full control over the feedback the system surfaces, which an end-to-end ML approach cannot provide.How Younalize uses CVAT OnlineYounalize’s training data comes from footage that co-founder Matt van Wezel captures during coaching sessions, with player consent, across indoor courts, beach courts, and training facilities in the Netherlands, Norway, and India. The team has accumulated between 5,000 and 7,000 training videos spanning a wide range of playing levels, lighting conditions, and ball colors. None of this footage comes from the app's users: Younalize's privacy policy explicitly commits that uploaded videos are never used for model training.The annotation pipeline is managed entirely through CVAT Online, and works in three stages:Frame extraction and local annotation. Python scripts run locally to extract frames from raw video clips, filter out unusable footage (poor angle, excessive motion blur, or frames where the player is out of frame), and run Younalize's custom YOLO and SAM 2 models to generate bounding boxes and pose labels. The complete pre-annotated batch is then uploaded to CVAT Online for verification.Verification in CVAT Online. Fabian reviews each batch of tasks in CVAT Online, accepting accurate pre-annotations and correcting misses, most commonly ball detections lost to motion blur or occlusion.Export, and retraining. Verified annotations are exported and combined with the corresponding frames to produce a new training dataset, which runs overnight on a dedicated GPU server. The next round of pre-annotations reflects the improved model.The loop runs continuously alongside Fabian and Matt’s other responsibilities. A batch of four or five video clips typically takes less than half an hour to verify. By the following morning, a new model version is ready.Why choose CVAT Online?Younalize initially ran CVAT Community, the open-source edition, on Fabian's laptop via Docker. The setup worked for solo annotation, but collaboration with Matt van Wezel proved impossible: copying project directories between machines failed repeatedly, and the two founders ended up annotating independently without a shared dataset. Moving to CVAT Online resolved the problem immediately: both founders could access the same projects, tasks, and job queues from wherever they were working.CVAT Online's API gives Younalize the flexibility to integrate their own custom models directly into the annotation pipeline as their workflow evolves. The team has used it to call their YOLO and SAM 2 models before frames reach the manual review stage, an approach that keeps human effort focused on the genuinely hard cases: motion-blurred ball detections, occluded keypoints, unusual playing conditions. The combined result is measurable: dataset preparation that previously took weeks now takes days. For a two-person team without dedicated annotation staff, that compression is the difference between shipping model improvements continuously and batching them into infrequent labeling sprints.1,300 users, 90 countries, no marketing budgetYounalize reached 1,300 active users across 90 countries without a paid marketing campaign. One national volleyball federation has already purchased 400 licenses for their pre-youth national training program. Another federation embeds Younalize in its annual coach development course. Each year, 100 to 150 trainers learn to run technique analyses on their players and design training plans from the feedback the app returns.The app is also being used as a remote coaching tool. A Dutch head coach working with an African national beach volleyball team uses Younalize to monitor player technique during the months he spends back in the Netherlands. Players upload their own clips; the coach reviews AI-generated feedback remotely and builds individualized training plans from the results.These use cases reflect the wider opportunity Younalize is pursuing. The International Volleyball Federation (FIVB) aims to grow the global volleyball movement from an estimated 800 million people today to 1.6 billion by 2032, according to its Strategic Vision 2032. The large majority of that growth will come from markets where professional coaching is unavailable or unaffordable. Younalize is in active conversations with the federation about embedding the app in their Volleyball Empowerment program, which places elite coaches in underserved markets to develop the sport from the ground up.The annotation workflow built on CVAT Online is already supporting this expansion. Field hockey is in active development, requiring new detection models for both the ball and the hockey stick. No reliable pre-trained models existed for either, so Younalize built them from scratch using the same pipeline. A strength and conditioning module is in progress for athletes in national training programs. Each new sport or technique follows the same path: collect footage, annotate in CVAT Online, train, iterate.Making expert coaching accessible at scaleYounalize is building the infrastructure to make expert sports coaching as accessible as a smartphone. The annotation pipeline that powers the AI, running on CVAT Online and driven by custom pre-annotation models managed by two founders, is what allows the product to improve continuously without a dedicated data team.CVAT Online provides the cloud access that makes collaboration possible, the API flexibility that lets Younalize's own models do the pre-annotation work, and the annotation environment where human judgment is applied only where it is genuinely needed. As Younalize scales across new techniques, new sports, and new markets, the same workflow scales with it.

The challenge of asbestos testing at scaleAsbestos is among the most heavily regulated materials in industry, and also one of the hardest to measure reliably. A single analyst at an accredited lab is expected to review hundreds of scanning electron microscope (SEM) images per sample, individually confirming the presence of fibers that are routinely micron-thick and sometimes as thin as 100 nanometers.The work is repetitive and unforgiving. Missing a fiber is not a quality issue — it is a compliance failure. And the volume of images makes consistent manual review nearly impossible to sustain.Scanning electron micrograph showing the distribution of suspected asbestos fibers and associated mineral fragments collected on a porous membrane substrate.Germany alone has roughly 100 asbestos-testing labs; Europe has around 300. Every one of them faces the same bottleneck: a human operator searching SEM tiles for something both hazardous and nearly invisible at scale.ProMetronics saw the same pattern across earlier custom AI projects — annotation workflows disconnected from training, pipelines that broke at handoffs, no unified environment to manage it all. Rather than solving those problems one client at a time, the team built a platform to eliminate them entirely. Asbestos detection became its first production use case.How CVAT fits into the ProMetronics pipelineThe ProMetronics pipeline starts when a lab sample arrives: the sample is prepared and loaded into a microscope, which captures around 1,000 images per job and up to 10,000 images per full sample batch. Those images are uploaded to ProMetronics' dataset hosting — and that's where CVAT comes in.Sparse-field labeling view in CVAT Online showing individual fiber annotations with polygon precision.Images are batched into projects and tasks inside CVAT Online. A dedicated annotation team works through the labeling load at polygon precision, while ProMetronics' in-house specialists run quality control using CVAT's built-in review-and-rework workflow. Once annotations pass review, they flow back into ProMetronics via the CVAT API and feed the training pipeline directly.ProMetronics doesn't build its own annotation UI. CVAT sits behind a thin pipeline layer inside the platform, meaning a lab customer never needs to learn or interact with it — they see dataset management, training, and inference, with annotations moving through in the background.Intense CVAT review session: annotation feedback comments surfaced inline across a dense fiber cluster.Beyond labeling, ProMetronics also uses the CVAT API as a live coordination signal. The German engineering team reads task status, job progress, QC state, and label counts directly from the API into their internal dashboard which gives them a real-time view over the annotation team's output without opening CVAT itself.Why choose CVAT Online?Three things kept ProMetronics on CVAT once they made the switch. The tool is productive out of the box, which matters when embedding it into someone else's platform rather than using it standalone. The review-and-rework stages are practical for a distributed team coordinating across time zones. And the API is robust enough to treat as infrastructure, not just a convenience, which is what ProMetronics needed to build its internal dashboard on top of."I love CVAT because it's so hands-on: working with it is easy to understand and easy to do. We tried other tools before, but we had a lot of friction getting them running and the UI was harder to work with."Andre Kempe, CEO & Product Owner, ProMetronicsBecause CVAT is the de facto annotation tool for computer vision teams, ProMetronics can also onboard new annotation partners quickly since they almost always already know the workflow.A working model, ready for the European marketThe results of the ProMetronics platform , with CVAT Online as its annotation backbone, are tangible. Approximately 100,000 asbestos fibers have been labeled at polygon precision across 30+ tasks and thousands of SEM images. A production Vision AI model has been trained, validated, and deployed into customer labs.Application interface of the ProMetronics Asbestos App showing active model inference. The integrated ProMetronics Vision AI model automatically identifies, segments, and highlights individual asbestos fibers using high-contrast green polygon overlays against a complex particulate background.For the labs using the platform, the end product is not an AI system. It is a turnkey workflow: a sample goes in, a compliance-ready report comes out, with fiber counts and dimensions — matching the format labs already use under regulatory pressure. The AI runs behind the electron microscope interface the operator already works with.The same pipeline generalizes beyond asbestos. The SEM → CVAT → training architecture applies to any domain that requires finding small, rare objects in high-resolution imagery — technical cleanliness testing, particle detection, and industrial quality control among them.What's next for ProMetronicsProMetronics' platform is production-ready and targeting the full European market: 100 testing labs in Germany and approximately 300 across the continent. The same annotation-to-inference pipeline that powered asbestos detection is already being extended to technical cleanliness and particle analysis workloads, with waveform and audio analytics on the roadmap.On the infrastructure side, the team continues to invest in the pipeline connecting CVAT to its training cluster — automating dataset expansion, streamlining model retraining, and improving the feedback loop between QC annotations and model performance over time.

‍The lake that wasn’t supposed to have themIn late 2022, Véronica Romero-Rosales and her partner David were visiting his parents at Lake Témiscouata, a long, cold lake in eastern Quebec. Over dinner they heard the news: zebra mussels — an invasive species most people associate with warmer waters like the Great Lakes — had turned up in a place no one expected.Lake Témiscouata, eastern Quebec — one of two recently invaded lakes Robonotic sampled to build its dataset.Véronica and David are both engineers, and both have a background in robotics and AI. The conversation kept circling the same question: if everyone knows the damage zebra mussels can do, why is nobody doing anything about it?“We were just thinking at the end of the day, how come nothing is being done? What can we do as citizens for the long term? That’s when we came up with the idea — the technology already exists. So it would be nice to create a solution that helps people monitor what’s going on underwater.”Véronica Romero-Rosales, Co-founder, RobonoticThe damage is not hypothetical. Forty years after infesting the Great Lakes, zebra mussels have thinned fish populations (that out-filter the zooplankton juveniles depend on), clogged municipal water pipes, and left beaches lined with razor-sharp dead shells. Véronica cites a small Canadian city (30,000 citizens) that just earmarked $1.2 million for a single year to keep its water intakes mussel-free.But that damage takes decades to compound. Catching an invasion early, when mussels are still hiding in isolated clusters, before the lakebed turns into a reef of shells, is the whole game. It is also the hardest moment to monitor, because there is almost nothing to see.Two engineers, one missionThe couple became Robonotic. Véronica took the CEO role; David took on hardware. But the project needed something neither engineer had at the start: a deep understanding of freshwater ecology.So Véronica went back to school. She enrolled in a master’s in oceanography under a professor who specializes in mollusks, pairing the biology with the AI. Along the way, Robonotic connected with two other organizations that became central to the project: the Quebec Artificial Intelligence Institute (Mila) and DeepSense, based in the Faculty of Computer Science at Dalhousie University. Mila helped the team think through model ethics and data strategy, while DeepSense matched the company with project managers and interns to help them develop their AI solution. “As engineers, we can do a lot of good things, but we can also do a lot of bad things when we don’t know the big picture. The biology part is very important. If we’re not careful enough, we can do more damage than good.”‍Véronica Romero-Rosales, Co-founder, RobonoticThat cautious posture shaped everything that came after — the partnerships, the data pipeline, the model, and the product roadmap.Learning to see underwaterRobonotic’s first break came from Bleu Massawippi, a Quebec association that has been fighting aquatic invasive species for more than six decades. Its general director, Laurence Renaud-Langevin, took the team under her wing; her divers became Robonotic’s first data-collection crew.But Véronica didn’t want to stay on the shore. “As a leader, you have to know exactly what are the efforts needed to get some data.” So she got her dive certification — which was harder than it sounds.Véronica on a data-collection dive in Lake Témiscouata.“I’m scared of the water, and I’m scared of small, confined places,” she says. “The tank is heavy and I’m small — five-foot-one.” Her first underwater attempts ended in anxiety; a patient instructor walked her through it step by step. Véronica now dives in Quebec summers — and the experience has reshaped how Robonotic scopes its own projects.“I have a really clear idea now of how many hours you have to plan ahead just to get one hour of footage,” she says. That clarity ends up mattering for every future commercial bid.A dataset built one frame at a timeRobonotic’s dataset comes from three deliberate sources.First, two recently invaded lakes — Lake Massawippi and Lake Témiscouata — rather than well-documented lakes already carpeted in mussel clusters. Robonotic specifically wanted footage of the hard cases: mussels hiding under rocks, small solitary ones, early colonies. A model trained only on dense clusters would be useless for the work the team actually wants to do.Second, a deliberate mix of cameras — GoPros for what a typical diver would carry in the field, plus an intermediate camera and two professional rigs for capturing the fine characteristics of juveniles and getting quality pictures even with turbidity for annotation. The variety means the model learns to recognize mussels regardless of the device the client happens to have on hand.One of Robonotic’s action-camera rigs: GoPro, dual LED lights, mounted on a dive-ready tray.Third, an improvised studio at an aquaculture station, where Véronica held roughly 200 live zebra mussels for six months. An aquarium, two underwater LED lights, a camera on a tripod — she shot thousands of controlled reference frames from every angle.Robonotic’s improvised photo studio at an aquaculture station: aquarium, underwater LED light, and a tripod-mounted camera.The reason for the variety: zebra mussels don’t look alike. “They all have different patterns in their shells. You might take a look at one and say, well, this doesn’t look like a zebra mussel — but it is one. The colors vary: blue, white, gray, all on the same lake.” A model trained only on textbook specimens would miss half of them in the wild.All told: roughly 52,000 frames, of which about 9,300 (~15%) are annotated and powering the current model. The team applies augmentation — blur, crops, rotations, color shifts — so the detector learns to cope with real-world underwater conditions: silt, low light, and motion blur.Why CVAT won outBefore landing on CVAT, Véronica evaluated three other tools. None of them was built for the way Robonotic actually works: multiple annotators, a central reviewer, a correction loop, and a project lead whose job is to keep everyone’s definition of “this is a zebra mussel” identical.“There wasn’t anything for working as a team,” she says of the alternatives. CVAT came up twice in parallel — once from her advisor at Mila, a Montreal-based artificial intelligence research institute, and again from the collaborating Dalhousie University team. She gave the CVAT Online Team plan a try and it stuck.“It was user-friendly and easy to learn. And since I knew I was going to be working with friends and colleagues who wanted to help out because they liked the project, I was looking for something easy to work with.”Véronica Romero-Rosales, Co-founder, RobonoticInside the annotation workflowVéronica runs annotation the way she used to run quality control in her previous engineering roles: tight SOPs, dummy-proof steps, and a real training ramp for every new annotator.Every volunteer, student, or friend who comes in reads the written work process first, then spends 30–60 minutes in a live training session before being assigned their first job. Inside CVAT, jobs move through explicit stages: annotation → verification → validation → done. Véronica uses verification herself to leave notes on individual bounding boxes (“tighten this box,” “wrong attribute,” “you missed one here”) and pushes jobs back for rework until they’re training-ready.Robonotic’s project in CVAT Online, with per-label annotation of zebra mussels and native freshwater mussels (“Mulette”).“We’re engineers — and I’ve worked in quality control before. Because we’re humans, we need dummy-proof processes just to have the same result every time. The stage and state workflow in CVAT is really helpful: I put notes, I send it back, and whenever a job is ready for training, we just switch the stage.”Véronica Romero-Rosales, Co-founder, RobonoticCompleted annotations export in YOLO format and feed training runs on AWS SageMaker, with experiments also running on an NVIDIA Jetson Orin — the same class of hardware Robonotic wants to eventually carry into the field for real-time, on-device detection.The team has also experimented with CVAT’s built-in SAM-based auto-labeling. It didn’t meaningfully accelerate their work yet — zebra mussels are small, textured objects that often blend into rocks, and Robonotic’s dataset is still small by foundation-model standards — but Véronica plans to try CVAT’s newer text-prompt segmentation and, down the road, bootstrap new annotations from Robonotic’s own fine-tuned model.“Someday I definitely want to use all the integrations,” she says. “The way we’ve been working wasn’t the most effective — but for the beginning, it was the right thing for us.”What kept them on CVATA lot of annotation tools could get the bounding boxes drawn. What kept Robonotic on CVAT, Véronica says, isn’t features — it’s support.“The reason I chose to stay with CVAT is because when I had issues, I got a quick answer from the support team. Nobody has time to lose. When you have to contact support at a larger tech company, it is hard to get an answer — but with CVAT, it’s easy and doesn’t take too much time. I felt there was a team behind the tool who was listening to the customer’s needs — willing to work with me, not just sell me something. The tool is great and user-friendly. But the support team behind it is what made the difference.”Véronica Romero-Rosales, Co-founder, RobonoticThat partnership cuts both ways. Because CVAT.ai also offers a managed labeling service to customers who need it, the same team working on the product is doing annotation projects themselves — and carrying those insights back into the tool.What’s next for RobonoticRobonotic is finishing its prototype phase and starting customer discovery in earnest. On the near-term list:The Quebec Ministry of the Environment, which has flagged roughly five lakes with environmental-DNA hits for zebra mussels that now need visual confirmation — the exact job Robonotic’s model is designed for.Continued work with Bleu Massawippi, Robonotic’s founding partner, whose monitoring program is where this story started.A presentation circuit — Montreal in May, El Salvador in October — aimed at aquaculture and freshwater-research communities. Véronica also speaks Spanish, and sees Latin America as a natural expansion market for the same technology pointed at different invasive species.“Our goal is to bring this kind of technology to normal biologists — people who don’t have the tools we have in other fields. To bring it into something that is very urgent for us to take care of.”Véronica Romero-Rosales, Co-founder, RobonoticNative freshwater mussels (“mulettes”) Robonotic is also learning to identify — species biologists want to protect, not remove.Robonotic’s current model already detects zebra mussels, invasive snails, and crayfish out of the same two-lake dataset. As the team takes on new waters and new clients, that list will grow. And as new jobs move through CVAT’s pipeline, the model gets sharper with them.

Challenge: A German developer of robotics hardware and software required an advanced point cloud data annotation suite in order to swiftly and accurately label sensor truck data for their autonomous trucking system. Result: In just five weeks, CVAT introduced several 3D cuboid labeling updates that enhanced precision, clarity, and speed for client’s 3D data annotation workflow.Why CVAT: The team chose CVAT Online Team plan for its essential labeling features, necessary for fast and accurate point cloud data annotation. They also chose it for its advanced quality and project management tools, which allow for efficient dataset distribution, annotation quality control, and annotator performance evaluation. FERNRIDE is a Munich-based deep-tech company specializing in autonomous, electric trucking solutions, particularly for logistics. It was spun out of the Technical University of Munich (TUM) in 2019 and focuses on "human-assisted autonomy," a hybrid approach where autonomous driving is combined with teleoperation, allowing human operators to remotely control trucks when needed while AI handles routine driving tasks.The company initially deployed driverless yard tractors and has expanded to serve major logistics operators, addressing challenges such as driver shortages, operational efficiency, and environmental sustainability. FERNRIDE’s technology is used in container terminals, logistics yards, manufacturing plants, and freight transport on private grounds. It aims to reduce costs, increase vehicle utilization, and decrease environmental impact by using electric trucks.Learn how FERNRIDE improves its autonomous truck vision with CVAT enhanced 3D labeling tools ChallengeTo turn conventional electric trucks into reliable autonomous assets [that can perceive their surroundings, make driving decisions, and execute transport tasks], FERNRIDE developed a platform that combines AI with cameras, sensors, and advanced software. Mounted on top of a vehicle, these systems not only enable real-time autonomy but also collect valuable visual data of what the trucks “see” during their rides. This data, in the form of point clouds and videos, is then labeled and fed back into the platform to continuously improve the vehicles’ perception, navigation, and decision-making capabilities on the road. However, labeling such data, especially 3D point clouds from LiDAR, is notoriously complex. While video annotation is relatively well-supported by many tools, 3D data introduces a steep learning curve and demands precise functionality to ensure annotation accuracy, especially for data captured in full-surround industrial settings.While FERNRIDE found CVAT’s 3D features promising in early tests, scaling up to larger and more complex datasets revealed several usability and workflow challenges:Limited precision when rotating cuboids: Adjusting object orientation was difficult due to fixed rotation angle, making it hard for annotators to align cuboids accurately in point cloud scenes.Inconsistent cuboid labeling: Adjusting cuboid shapes and correcting errors took too many steps, leading to annotation inconsistency.Limited support for object relationships and visibility: Understanding spatial relationships (e.g., cargo stacked on pallets, vehicles crossing paths) was difficult without more advanced visual cues and orientation tools.These limitations hindered the team's progress in labeling and training AI models.SolutionRecognizing the limitations, the CVAT team stepped in to address them right away. For every issue the customer surfaced, CVAT allocated dedicated engineering resources to ensure timely, targeted updates aligned closely with FERNRIDE’s expectations.As a result, within just 5 weeks, the CVAT team released three major feature updates specifically designed for more precise and intuitive point cloud labeling.1. Smooth Cuboid Rotation for Accurate Point-Cloud and Label AlignmentFERNRIDE's first bottleneck was adjusting the orientation of 3D cuboids when labeling objects in LiDAR scenes. The fixed rotation speed in CVAT’s 3D workspace made it difficult to fine-tune angles, particularly when annotating objects that required precise alignment, such as cargo containers or vehicles in tight spaces. This, in turn, slowed down the annotation process, and the labeling quality.To address this, CVAT introduced dynamic 3D cuboid rotation that adapts to the annotator’s mouse movement. Now, faster mouse movement results in a larger rotation angle, allowing for quick adjustments. Slower movement yields finer control, allowing annotators to focus on details without overshooting.2. Precise Cuboid Size Input Controls for Centimeter-Level PrecisionThe second update addressed the size of the cuboids used to label objects in 3D scenes. FERNRIDE's machine learning team required precise annotations down to the centimeter in order to train systems to reliably handle cargo containers, pallets, and barriers. Without this level of precision, annotators had to "eyeball" object dimensions, which resulted in inaccurate and inconsistent annotations, and inaccurate perception by their AI vision models.To solve this problem, CVAT introduced manual cuboid input fields that gave annotators full control over the exact height, width, and depth measurements of each 3D object. Since container dimensions are typically standardized and known in advance, this feature allowed annotators to label identical containers consistently across multiple scans instead of estimating their size visually. As a result, labeling became both faster and more reliable, improving data quality and model performance for tasks requiring high spatial accuracy.3. Cuboid Orientation Arrows for Directional AccuracyThe final, yet important update was related to object orientation in LiDAR scenes. In container terminals and logistics environments, misjudging the direction of trucks, cargo, or barriers can lead to significant risks both in model training and in real-world deployment. FERNRIDE’s annotation team struggled to consistently align cuboids with the true facing direction of objects in a point cloud environment. Without a clear directional indicator, some cuboids were rotated incorrectly introducing noise into the dataset and affecting motion tracking models downstream.To overcome this, CVAT introduced orientation arrows that allow annotators to quickly and accurately set the direction of labeled objects. These arrows dynamically represent the cuboid’s orientation along the X, Y, and Z axes across all views, including Perspective, Top, Side, and Front. They are scalable based on cuboid size, configurable to suit annotation needs, and can be toggled on or off for a cleaner workspace.ResultsThe impact of CVAT’s 3D enhancements, shaped directly by FERNRIDE’s feedback, was felt across every stage of the client’s labeling workflow. With the ability to precisely control cuboid dimensions, accurately set object orientation, and adjust cuboid positions more smoothly within dense point clouds, FERNRIDE unlocked several key benefits:1. Greater annotation accuracyCentimeter-level control over cuboid size and orientation significantly reduced labeling errors, resulting in cleaner datasets and more reliable training inputs for their perception models.2. Faster labeling workflowsEnhanced 3D controls and smoother cuboid manipulation allowed annotators to adjust object dimensions and orientations more intuitively in dense point cloud scenes. This streamlined the labeling process, accelerating project delivery timelines by 30% and reducing time-to-deployment for FERNRIDE’s perception models.3. Stronger AI models and broader applicationsHigher-quality point cloud annotations directly improved the performance of FERNRIDE’s AI models, enabling more accurate motion tracking, better scene understanding, and safer autonomous vehicle behavior in real-world logistics environments.Going forward, insights from FERNRIDE’s use cases will continue to shape CVAT’s roadmap for scalable, high-precision 3D labeling.“Working with FERNRIDE showed how deep collaboration with customers can directly shape the evolution of our platform. The 3D features we built together are already making annotation easier and more precise for other teams tackling complex, high-stakes use cases in autonomous driving, robotics, and industrial automation,” said Nikita Manovich, CEO of CVAT.