Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/RohanDoshi2018/ZeroshotSemanticSegmentation
Zeroshot learning for semantic segmentation using joint visual-semantic embedding space.
https://github.com/RohanDoshi2018/ZeroshotSemanticSegmentation
Last synced: 14 days ago
JSON representation
Zeroshot learning for semantic segmentation using joint visual-semantic embedding space.
- Host: GitHub
- URL: https://github.com/RohanDoshi2018/ZeroshotSemanticSegmentation
- Owner: RohanDoshi2018
- Created: 2018-01-23T03:31:41.000Z (almost 7 years ago)
- Default Branch: master
- Last Pushed: 2018-07-30T18:37:02.000Z (over 6 years ago)
- Last Synced: 2024-08-01T20:36:39.995Z (3 months ago)
- Language: Python
- Homepage:
- Size: 9.6 MB
- Stars: 49
- Watchers: 3
- Forks: 5
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-computer-vision-papers - 2019NIPS
README
# Zero-Shot Semantic Segmentation
>
- Princeton University Senior Thesis
- Advisor: Professor Olga Russakovsky, Princeton University Department of Computer Science
- Part of the [Princeton VisualAI Lab](http://visualai.princeton.edu/people.html)## Project Overview
This is the PyTorch implementation of the seenmask zeroshot network (SZN) described in Rohan Doshi's senior thesis "Zero-shot Semantic Segmentation." Please reference this paper (rohan_doshi_senior_thesis.pdf) to understand the code.
## Installation
** Requirements: ** Conda (with Python 3) and Linux
1. Install Conda
2. Clone repository
```bash
git clone https://github.com/RohanDoshi2018/ZeroshotSemanticSegmentation.git
cd ZeroshotSemanticSegmentation
```3. Create new conda environment
```bash
conda create --name thesis_env
```4. Install Dependencies
```bash
conda install pytorch torchvision -c pytorch
pip install pytz pyyaml scipy fcn jupyter tensorflow tensorboardX
```5. Activate your conda environment
source activate thesis_env6. Run code
```
./train.py -c 4 -g 0
```7. [Optional] Run Tensorboard Server. Use Ngrok tunnel to access server remotely.
```
tensorboard --logdir /opt/visualai/rkdoshi/ZeroshotSemanticSegmentation/tb
```