Video object detection using Visual AI¶

Access this AI accelerator on GitHub

Object detection (binary and multiclass classification) applied to image and video processing is one of the tasks that can be easily and efficiently implemented with DataRobot Visual AI, which allows you to train deep learning models intended for Computer Vision based-projects. You can also bring your own Computer Vision model and deploy it in DataRobot via the Custom Model Workshop.

This accelerator demonstrates how deep learning modela trained and deployed with DataRobot can be used for object detection on a video stream. (Consider the example of detection when the person in front of the camera wears glasses.) The Elastic-Net Classifier (L2 / Binomial Deviance) along with Pretrained MobileNetV3-Small-Pruned Multi-Level Global Average Pooling Image Featurizer with no image augmentation are used in this accelerator. The dataset used contains images for two classes: persons with glasses and persons without glasses and is linked in the accelerator on GitHub. A sample of the dataset (100 images for each class) is used for this accelerator. The video stream is captured with OpenCV Computer Vision library. The frontend is implemented as a Streamlit application.