From a2222db8b6d24e1b872c820995b919b46fade7a6 Mon Sep 17 00:00:00 2001 From: Pang Date: Wed, 5 Jun 2024 19:16:39 +0800 Subject: [PATCH] Update README.md --- README.md | 34 +++++++++++++++++++++++++++++++++- 1 file changed, 33 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 0b09892..15b56b9 100644 --- a/README.md +++ b/README.md @@ -1,2 +1,34 @@ # OVCamo -Open-Vocabulary Camouflaged Object Segmentation + +Related Paper: Open-Vocabulary Camouflaged Object Segmentation + +## Dataset Details + +### Prepare Related Data + +- Collected by Fan *et al.*: + - **COD10K**, **CHAMELEON**: +- **PlantCamo**: +- **CPD1K**: +- Annotated and Generated by Cheng *et al.*: + - **MoCA-Mask**: + - **CAD**: + - Mask Generated by SLT-Net: + +### Data Usage + +1. Unzip all the files and set the following path information in the project's `.env`: + - `COD10K_TRAIN_DIR`: Directory of the train set of **COD10K**, which contains `Imgs` and `GT` in `COD-TrainDataset`. + - `COD10K_TEST_DIR`: Directory of the test set of **COD10K**, which is the path of `COD10K` in `COD-TestDataset`. + - `CHAMELEON_DIR`: Directory of **CHAMELEON**, which is the path of `CHAMELEON` in `TestDataset_per_sq`. + - `PlantCamo_DIR`: Directory of **PlantCamo**, which contains `train` and `test` directories of **PlantCamo**. + - `CPD1K_DIR`: Directory of **CPD1K**, which contains `img` and `gt` of CPD1K. + - `MoCA-Mask_DIR`: Directory of **MoCA-Mask**, which contains `TrainDataset_per_sq` and `TestDataset_per_sq`. + - `CAD_IMAGE_DIR`: Directory of **CAD** images, which contains directories of all sequences. + - `CAD_MASK_DIR`: Directory of **CAD** masks generated by SLT-Net, which contains directories of all sequences. +2. Specify the data roots for training and testing, and run the script to split the dataset: `python split_dataset.py --data-json ovcamo.json --train_root TRAIN_ROOT --test_root TEST_ROOT` +3. Use the training and testing splits of dataset in the training and inference process. + +## LICENSE of OVCamo + +

OVCamo by Youwei Pang, Xiaoqi Zhao, Jiaming Zuo, Lihe Zhang, Huchuan Lu is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International