Ultralytics HUB web - Vision AI training center
Ultralytics HUB is a Vision AI platform where you can visualize, train and deploy all your YOLOv5 models for free.
The users can visualize their data, train AI models, and deploy them to the real world in a seamless experience brought to them by the creators of YOLOv5.
How does Ultralytics HUB work?
Step 1: Upload your data
Your goal is to feed your model with quality data for it to recognize the characteristics that make up the object you want it to identify. Doing so is simple. All you have to do is upload a dataset of images of the object you want your model to understand. I.e., to train a model to recognize whether a person is wearing their mask properly, simply upload images of people wearing masks.
Step 2: Train your model in just a few clicks
Your model needs to train for the same reason as an athlete: to maximize performance. The only difference is that training your model in Ultralytics HUB requires no heavy lifting on your end. You only need to do 3 things here:
Select your dataset
Select one of our pre-trained YOLOv5 models
Connect to an agent and let Ultralytics HUB do the rest.
Step 3: Deploy your model to the real world
Congrats, you’ve trained your model! It now has sufficient knowledge and is ready to be deployed in the real world. From Ultralytics HUB, you’ll have the flexibility to export your model to 11 different formats quickly.
Simply download your model to your preferred format according to your device and preview it in action!
Source
The first month I dedicated myself to only developing the missing screens, providing consistency to all the design system elements, and applying a UI based on the user profile that according to the company we were targeting.
After the launch of the first MVP, the company decided to hire me full-time, so we planned a second launch of the platform, now with a little more time but with the same amount of resources.
The first thing we did was a thorough research job to see how platforms similar to ours worked. Lists of features were made and it was studied how they approached the most common actions, analyzing user journeys, seeing for example how many clicks were made on each platform to train a model.
Together with the project management, we created the first userflow for the web and app HUB, while Marketing took care of creating the user personas, and the programmer placed Hotjar within the web platform, with the idea of obtaining feedback from the users.
Once the beta version was online, we invite the first 200 users to log in and test all the HUB functionalities with the aim of obtaining valuable feedback.
After analyzing the bugs found and the UX problems in which the users were stuck, we decided to modify and improve the training dashboards and the way in which certain functions were accessed.
Studying the heat maps and the videos of the users' sessions, we saw that they constantly clicked on the Ultralytics logo, we later understood that we should move forward in the creation of a dashboard where we can place charts, notifications and a summary of the activity of the user on the platform.
As time went by, the platform grew in terms of users and feedback. We reviewed the userflow several times and modified it as we developed new features. The programmers were able to read the training information of the models in real-time and we placed new charts. We also take advantage of the letter system to make a more modular Dashboard that we can also use to promote our Newsletter and our social networks.
The evolution of Ultralytics HUB's user flow has been a dynamic journey, driven by a deep understanding of user behavior and an unwavering commitment to enhancing the platform's functionality and user experience. Our insights were gleaned from the analysis of heatmaps and user session videos, revealing a consistent pattern of users clicking on the Ultralytics logo. This valuable observation prompted us to embark on the creation of a dedicated dashboard, designed to house informative charts, notifications, and a user activity summary.
As time progressed, our platform expanded in both user base and feedback volume. With each growth milestone, we revisited and refined the user flow, adapting it to accommodate new features and user requirements. One significant enhancement was the implementation of real-time access to model training data for programmers, allowing for instantaneous insights and feedback. We integrated additional charts to provide users with a comprehensive overview of their model's progress.
Additionally, we leveraged a modular dashboard system, capitalizing on the versatility of card-based layouts. This not only empowered us to seamlessly introduce new features but also facilitated the promotion of our Newsletter and social media channels to foster community engagement.
Today, Ultralytics HUB boasts a thriving user community of over 15,000 users. We remain dedicated to refining the platform further, particularly by bolstering the backend infrastructure to expedite the Vision AI model training process. Our ongoing mission revolves around making this process faster, more precise, and simpler, always with a sharp focus on the needs and expectations of our valued users. The journey continues, driven by our commitment to delivering a cutting-edge Vision AI platform that empowers users to achieve their goals with ease and efficiency.
