Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,58 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
- pytorch
|
| 5 |
+
- image-classification
|
| 6 |
+
- gzsl
|
| 7 |
+
- agriculture
|
| 8 |
+
- weeds
|
| 9 |
+
- crops
|
| 10 |
+
- mobilenetv2
|
| 11 |
+
- resnet18
|
| 12 |
+
- squeezenet
|
| 13 |
+
- shufflenetv2
|
| 14 |
+
- squeeze-and-excitation
|
| 15 |
+
- depthwise-separable-convolution
|
| 16 |
+
- weed-identification
|
| 17 |
+
---
|
| 18 |
+
|
| 19 |
+
# GZSL Weeds Identification: Lightweight Classifier Weights
|
| 20 |
+
|
| 21 |
+
This repository hosts the PyTorch checkpoints used in our generalized zero‑shot learning (GZSL) pipeline for weed identification in agricultural imagery.
|
| 22 |
+
Backbones were fine‑tuned on **CropAndWeed** and evaluated for cross‑dataset generalization to **Plant Phenotyping** and a self‑collected, real‑field dataset.
|
| 23 |
+
|
| 24 |
+
## Available models
|
| 25 |
+
|
| 26 |
+
| File name | Architecture / variant |
|
| 27 |
+
|-----------|------------------------|
|
| 28 |
+
| `mobilenet.pt` | MobileNetV2 (ImageNet stem, width 1.0) |
|
| 29 |
+
| `resnet18.pt` | ResNet‑18 |
|
| 30 |
+
| `squeezenet.pt` | SqueezeNet 1.1 |
|
| 31 |
+
| `shufflenet.pt` | ShuffleNet V2 (baseline) |
|
| 32 |
+
| `shufflenet_squeeze_excitation.pt` | ShuffleNet V2 + Squeeze‑and‑Excitation (SE) |
|
| 33 |
+
| `shufflenet_sep_conv.pt` | ShuffleNet V2 + Depthwise Separable Convolution (SC) |
|
| 34 |
+
| `shufflenet_sep_conv_squeeze_excitation.pt` | ShuffleNet V2 + SC + SE |
|
| 35 |
+
|
| 36 |
+
## Getting started
|
| 37 |
+
|
| 38 |
+
All inference scripts, data loaders and architecture definitions live in the companion GitHub repository:
|
| 39 |
+
https://github.com/SyArsRa/WeedZSL.git
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
The quick‑start guide there walks through:
|
| 43 |
+
|
| 44 |
+
1. Instantiating the desired backbone (for example, MobileNetV2 or ShuffleNetV2 + SE)
|
| 45 |
+
2. Loading the matching `.pt` file from this weights hub
|
| 46 |
+
3. Running single‑image or batch inference
|
| 47 |
+
4. Fine‑tuning on a custom dataset if needed
|
| 48 |
+
|
| 49 |
+
## License
|
| 50 |
+
|
| 51 |
+
Weights and code are released under the MIT license for research and non‑commercial use.
|
| 52 |
+
See `LICENSE` for details or contact the maintainers for alternative licensing.
|
| 53 |
+
|
| 54 |
+
## Citation
|
| 55 |
+
|
| 56 |
+
These checkpoints support a study currently submitted to AAAI 2026.
|
| 57 |
+
Please cite the forthcoming paper or contact the authors for an interim reference.
|
| 58 |
+
|