Vision AEye cloud is a cloud based platform where users can create events and annotate images with bounding box, polygons or key-point annotations. The platform allows users to train and deploy AI models either on cloud or on your on-premise server. This is a complete platform for enterprise level vision AI projects.

You need the Vision AEye On-prem software to connect cameras and run the AI on your on-prem server. The software gets installed within your on-prem environment (preferably Linux environment) and automatically connects Vision AEye cloud to download trained AI models within the on-prem environment. It sends the live streams from cameras to the AI model for detection and alert/report.

Stream Sender is a software that sends video streams from one network to another network. For enterprises who have multiple locations and networks with hundreds of cameras need the Stream Sender software to connect cameras from different networks to a on-prem or cloud server for video processing. Install the software in any machine within the network where your cameras are located. You need to connect your cameras using RTSP link to this software and generate a unique key called Stream Key. Use these stream keys to the Vision AEye On-prem to connect these cameras that are located within a different network.

Vision AEye cloud and Vision AEye On-prem work together seamlessly. They are alway connected with each other. You can send live camera streams to Vision AEye cloud server for real-time video processing. Similarly you can automatically get your AI model that you have trained within the Vision Cloud to the Vision AEye On-prem. During annotation or feedback loop, you can send images captured by your camera to the Vision AEye cloud for the AI model finetuning and training.

Event is an incident or object that you want to detect within a video stream. For Machine Inspection and defect detection, Product defect is an event. Similarly, for colleges, fighting is an event. Within the Vision AEye Cloud, you can create an event and within that event you can label images and train the AI model. The Event that you have created within the Vision AEye cloud will be visible within the Vision AEye On-prem. The AI model that you have trained within the Vision AEye cloud will be automatically deployed to the Vision AEye On-prem under the same event. Now, simply add cameras to that event and start detecting objects/events.

Launch the Vision AEye on-prem and go to the event that you have created or you have within your platform. From settings, add cameras for each event. You can add multiple cameras within one event. If your cameras are located in a different network, use “Stream Key” generated by Intelgic’s Stream Sender software to connect those cameras.

We are using the reinforcement learning algorithm within our annotation tool. This will enable you to annotate only a few images and the algorithm automatically starts detecting the object within new images even during the annotation. It will also create bounding boxes or polygons on its own. Within minutes you can annotate and prepare a dataset for the AI training.

After you have a trained model, our system will create an API and webhook for each event. Use this webhook and API to integrate other systems to act or generate your own reports based on your requirements.

Train your own AI model with fewer images using the state of the art annotation tool and deploy it either on your on-premises server or on the cloud.