AI For Custom Object Detection & Image Processing

What Is Image Processing & Training?

Data Collection

The first and foremost step is to collect relevant image data to develop accurate artificially intelligent solutions.

Labeling

Labeling is the process of identifying and manually marking the intended objects like car, bus, person, etc visible in the images.

Training

Training is the process of creating AI models with labelled datasets. The bigger and more accurate the data set, the better the training will be.

Auto Object Detection

Finally, after repeated training, the trained model can auto-detect the target objects.

OpticVyu Custom Object Detection Process

OpticVyu's artificial intelligence-based image processing solution helps you collect the raw image data, allows labelling on the web dashboard, trains the model for your custom dataset, and detects the custom objects. Here is the complete process explained with the use-case of detecting construction workers and equipment:

Install OpticVyu HD Cameras & Start Collecting the Image Data

The first step is to install an OpticVyu HD camera at your job site. The OpticVyu team will help in identifying the perfect camera installation location. After successful installation, the camera will start to capture and send the images directly to the cloud-based dashboard. Here, image data will be stored for labelling and training.

Start Labeling

Once the collected image data is sufficient for training, the OpticVyu team will label or annotate the images containing the target objects. For example, in construction images, the OpticVyu team labels the workers, excavators, telehandlers, cranes, trucks, etc (as shown in the image). The labeling process continues until the dataset is considerable enough for the training. The labeled images can be checked anytime through the OpticVyu web based dashboard.

Training Of The Dataset

As the labelling dataset grows, the OpticVyu team repeatedly performs the computer vision training using its High-end NVIDIA Tesla GPUs based dedicated cloud server instances. The training process continues until the custom object detection accuracy reaches more than 75%. The resulting "artificial intelligence model" is a Masterpiece AI Model for the organisation that will be able to automatically recognise the target objects in any new image file.

Install Cameras To Other Jobsites And Auto-detect Custom Objects

The Masterpiece AI Model is now ready to detect the custom objects on your project sites. Now install the OpticVyu HD cameras wherever you want to detect the custom objects. Our team will help you with the camera installation and setting up the AI detection.

Visualize the Data & Generate Reports

The OpticVyu dashboard features graphical representations to visualize the custom object detection data for reconciling past data and making future decisions. Also, you will be able to generate custom reports for the custom object detection data.