DENGRENHAO / CSAILVision_semantic_segmentation

Semantic Segmentation implementation from CSAILVision

Semantic Segmentation

Getting all 150 classes pixel percentage from semantic segmentation. Using pretrained HRNet on MIT ADE20K dataset, most of the code taken from CSAILVision and simple inference implementation from liuch37.

Prerequisite

git clone https://github.com/DENGRENHAO/CSAILVision_semantic_segmentation.git

cd .\CSAILVision_semantic_segmentation\

pip install -r requirements.txt

Usage

範例

python semantic_segmentation.py -i ./input_img_folder/ -o ./output_img_folder/

• 在 ./input_img_folder/資料夾中的圖片皆會經過Semantic Segmentation，最後輸出結果至./output_img_folder/資料夾中

Results

semantic map