ShenXinda / ORBSLAM2_Semantic_Mapping

Semantic mapping based on pixel level classification.

ORBSLAM2_Semantic_Mapping

Semantic mapping based on pixel level classification.

1. Reference

• ORB-SLAM2
• deeplab-pytorch

2. Prerequisites

The weight file should be placed in deeplabv2/data/models，and downloads in deeplab-pytorch . And you can change the path in deeplabv2/inference.py.

class SegModel():
    def __init__(self):
        config_path = "deeplabv2/configs/cocostuff164k.yaml"
        model_path = "deeplabv2/data/models/deeplabv2_resnet101_msc-cocostuff164k-100000.pth"
        ......

3. Build and Run

The same as ORB-SLAM2.

4. Explain (Chinese)

Refer to 基于ORB-SLAM2的语义地图构建.