DeepLabV3+ extends atrous convolution–based semantic segmentation with an encoder–decoder structure that refines object boundaries while preserving rich contextual information.

Original paper: Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation (DeepLabV3+)

DeepLabV3Plus-ResNet50

This model uses DeepLabV3+ with a ResNet-50 backbone, combining multi-scale context aggregation from atrous spatial pyramid pooling (ASPP) with a lightweight decoder for sharper segmentation outputs. It is well suited for semantic segmentation tasks in applications such as autonomous driving, robotics, and scene understanding, where accuracy and robustness are critical.

Model Configuration:

• Reference implementation: DeepLabV3Plus_ResNet50_v1
• Original Weight: DeepLabV3Plus_ResNet50_Weights.Pascal_VOC2012_Aug
• Resolution: 3x513x513
• Support Cooper version:
  • Cooper SDK: [2.5.2]
  • Cooper Foundry: [2.2]

Model	Device	Model Link
DeepLabV3Plus-ResNet50	N1-655	Model_Link
DeepLabV3Plus-ResNet50	CV72	Model_Link
DeepLabV3Plus-ResNet50	CV75	Model_Link