Skip to content

Latest commit

 

History

History
42 lines (26 loc) · 1.22 KB

README.md

File metadata and controls

42 lines (26 loc) · 1.22 KB

Place classification using DeepLab-v2 with PyTorch

This is an implementation of place classification method of "Place classification based on semantic segmented objects".

COCO-Stuff

Pre-trained Deeplab-v2 with COCO-stuff 164k dataset is available at 164k Model.

Setup

Requirements

  • Python 2.7+/3.6+
  • Anaconda environment

Then setup from conda_env.yaml. Please modify cuda option as needed (default: cudatoolkit=10.0)

$ conda env create -f configs/conda_env.yaml
$ conda activate deeplab-pytorch

Datasets

Setup instruction is provided in each link.

Building weight matrix

Usage: place_weight_matrix.py single -c ./configs/cocostuff164k.yaml -m ./models/deeplabv2_resnet101_msc-cocostuff164k-100000.pth -i .

Place classification using built weight matrix

Usage: place_classification.py single -c ./configs/cocostuff164k.yaml -m ./models/deeplabv2_resnet101_msc-cocostuff164k-100000.pth -i .

References