The Chooch AI Platform is built for AI training and deployment, including data collection, AI model management, and ultimately creating real world computer vision programs that get results.
At the core of the platform, the Chooch AI Dashboard imports visual data, provides annotation and labelling tools, trains the models, and then exports AI models for the cloud or edge AI.
Customers use the Chooch AI dashboard for AI training, and also rely on our team to create bespoke models, annotating or labeling data sets, training the AI to identify the objects, actions, images to automate visual tasks.
Chooch AI has also created standard pre-trained AI models for many common tasks – from safety equipment detection to visual fire alerts.
From inventory tracking to facial authentication, from action detection to people counting, Chooch AI provides complete solutions that reduce risk, reduce costs and increase productivity for any industry.
Here is an overview of several types of AI training, which clarify just how easy it is to use the Chooch platform. Videos are also available to walk anyone at any level of learning through the process, novice to expert .
Chooch offers a powerful, mature AI platform that automates the process of AI training without the need to hire expensive in-house developers.
The dashboard offers annotation and labelling, including video annotation that can generate hundreds of annotated images simultaneously. These tools unblock bottlenecks in AI training.
Training is complete when high enough accuracies are achieved for your particular use case. Even then, improvements can be made through the dashboard for greater accuracy, and revised AI models are deployed to the cloud or the edge automatically.
AI models in Chooch are known as “Perceptions,” each one containing one or more classes.
For example, you might create a Perception for recognizing brand logos, and then create a separate class for each logo (e.g. Nike, McDonald’s, etc.).
The first step in training an AI model is to upload your images or videos to the Chooch dashboard, which is technically called data collection.
Next, each image needs to be labeled (if performing image classification) or annotated (if performing object detection). Annotations may take the form of bounding boxes or polygonal annotations (as described above).
Finally, AI training can proceed, and AI models are generated.
When videos are used to do AI training on the Chooch AI dashboard, the platform separates videos into individual frames that are annotated just like images.
After uploading a video to the dashboard, you draw bounding boxes around objects that you want to train on.
Chooch then automatically generates annotated training data by following the motion of these objects throughout the video.
If necessary, you can also manually edit the dataset once it’s been generated to add or change the annotations.
This type of AI training creates datasets very rapidly.
Facial authentication is a special case of object recognition that seeks to recognize different human faces for verification, for access to spaces or data, to check in, or for other security measures.
As with image classification, training an AI model for facial authentication starts by creating a new Perception.
For example, you might create a Perception for the list of employees allowed in a restricted area, where each class in the Perception represents a different employee, either anonymized or known.