These codes and datsets are from the paper "Growing status observation for oil palm tree using Unmanned Aerial Vehicle (UAV) images", which is under review in ISPRS Photogrammetry and Remote Sensing (Major Revision).
CUDA_VISIBLE_DEVICES=gpu_id python tools/train.py configs/oilPalmUav/mopad.py
CUDA_VISIBLE_DEVICES=gpu_id python demo/demoFull.py configs/oilPalmUav/mopad.py work_dirs/mopad/latest.pth mopad-det.txt test_images
Our training models for Site 2 can be downloaded from
Baidu Wangpan Access: 7n61
Our training models for Site 1 can be downloaded from
Baidu Wangpan Access: 8mwa
Our dataset for Site 2 can be downloaded from
Baidu Wangpan Access: qpaw
Our dataset for Site 1 can be downloaded from
Baidu Wangpan Access: fgfv
The data should be saved in the folder ./data
We followed COCO format basically.
The structure of the dataset is as follows:
train2017
: images for training dataset (like<id>.jpg
)val2017
: images for validation dataset (like<id>.jpg
)annotations
: annotations includinginstances_train2017.json
andinstances_val2017.json
for training and validation dataset, respectively