weecology / DeepTreeAttention

Implementation of Hang et al. 2020 "Hyperspectral Image Classification with Attention Aided CNNs" for tree species prediction

MIT License

119 stars 38 forks source link

LiDAR arm #11

Closed bw4sz closed 3 years ago

bw4sz commented 4 years ago

Add LiDAR data. Needs to be fleshed.

bw4sz commented 4 years ago

Data Generation

Decide on 3d point cloud versus CHM. Probably CHM to start with.
[ ] Create a lidar.py to crop and generate CHM
[ ] add to boxes.py generate tfrecords for weak label pipeline https://github.com/weecology/DeepTreeAttention/blob/63d7025a83385b775996b2e46212eb2280dd68ac/DeepTreeAttention/generators/boxes.py#L24
[ ] add to prepare_field_data.py to add to NEON label pipeline

Model

[ ] Create lidar_CNN.py model in https://github.com/weecology/DeepTreeAttention/tree/master/DeepTreeAttention/models. Cannot see current model since it has only spatial structure, probably just use spatial arm here

https://github.com/weecology/DeepTreeAttention/blob/63d7025a83385b775996b2e46212eb2280dd68ac/DeepTreeAttention/models/layers.py#L24

bw4sz commented 3 years ago

Added height to metadata, enough for now.