はてなブックマーク

Region of Interest pooling (RoI pooling) is a widely used technique in object detection with convolutional neural networks. It extracts fixed-size feature maps (e.g., 7×7) from variable-sized regions, enabling efficient classification of multiple objects like cars or pedestrians. RoI Align refines this process by avoiding coordinate quantization and using bilinear interpolation, resulting in more

We resized and aligned 16-band channels to match those from 3-band channels. Alignment was necessary to remove shifts between channels. Finally all channels were concatenated into single 20-channels input image. Model Our fully convolutional model was inspired by the family of U-Net architectures, where low-level feature maps are combined with higher-level ones, which enables precise localization.

Right Whale Recognition was a computer vision competition organized by the NOAA Fisheries on the Kaggle.com data science platform. Our machine learning team at deepsense.ai has finished 1st! In this post we describe our solution. The challenge The goal of the competition was to recognize individual right whales in photographs taken during aerial surveys. When visualizing the scenario, do not forge

Explore all the technology expertise we have to develop AI solutions