license: gpl-3.0
pipeline_tag: image-segmentation
tags:
- AIoT
- QNN
U-Net: Semantic Segmentation
U-Net is a convolutional neural network designed for biomedical image segmentation, introduced by Olaf Ronneberger et al. in 2015. The model gets its name from its U-shaped architecture, featuring a symmetrical encoder-decoder structure. The encoder part extracts features from the image through a series of convolutions and downsampling operations, while the decoder part restores the spatial resolution through upsampling, combining the extracted features to accurately locate and segment objects within the image. U-Net uses skip connections that pass feature maps from the encoder directly to the decoder, aiding in the recovery of fine details. This design makes U-Net highly effective for tasks requiring precise localization, such as medical image segmentation, and it is widely applied in other areas like remote sensing, autonomous driving, and image denoising.
Source model
- Input shape: 640x1280
- Number of parameters: 29.6M
- Model size: 118.4M
- Output shape: 1x2x640x1280
Source model repository: U-Net
Performance Reference
Please search model by model name in Model Farm
Inference & Model Conversion
Please search model by model name in Model Farm