Posts
-
Binary Image Segmentation Using Synthetic Datasets
In computer vision, one of the most common challenges is to remove the background from the foreground of an image. A popular solution to this problem is to use a semantic segmentation model to separate the foreground from the rest of the image. There are several general-purpose models available that can solve the binary segmentation problem, such as U-Net, Dichotomous Image Segmentation (DIS), Mask R-CNN, and Fully Convolutional Networks. -
A Quantitative Comparison of Serving Platforms for Neural Networks, part 2
We tested REST APIs of TensorFlow serving, TorchServe, and NVIDIA Triton Inference server in the previous post. In this article, we will look at gRPC APIs. We measured and compared response time, request rate, and failure amount depending on the load. -
A Quantitative Comparison of Serving Platforms for Neural Networks
Choosing the suitable method of production serving your neural network model is one of the most critical decisions. We tried to compare the most popular serving platforms performance, stability and usage complexity.
subscribe via RSS