Problem: Get a deep learning neural network model to identify objects in photos.
Context: Image classification (recognition) is one of the foremost capabilities of deep neural networks. Inception-v3 is one of the most popular convolutional neural network models for recognizing objects in images. Deep learning-powered image recognition is used by doctors to identify cancerous tissue in medical images, self-driving cars to spot road hazards, and Facebook to help users with photo tagging.
Model type: Deep convolutional neural networks model (Inception-v3)
What we did: We deployed an Inception-v3 model using a SnapLogic Ultra Pipeline, a powerful, low-latency data pipeline. (More on how we built this demo.) The model was trained on an ImageNet dataset containing 1,000 types of objects. SnapLogic provides both horizontal and vertical scale to deep learning models and supports GPU acceleration.
In this demo, take a photo of an object (e.g., a coffee mug, a candle, a sweater, etc.) using your webcam. (We do not store your images from this demo.) The photo will then go to the SnapLogic Ultra Pipeline, where the deep learning model will analyze the image. The model will then yield an output in which it tells you what object it perceives in the image and how confident it is in its interpretation. This process takes a few seconds, but it can be accelerated with GPUs if needed.