Upload inference.py
Browse files- inference.py +19 -1
inference.py
CHANGED
|
@@ -1,5 +1,23 @@
|
|
| 1 |
"""
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
"""
|
| 4 |
|
| 5 |
from __future__ import annotations
|
|
|
|
| 1 |
"""
|
| 2 |
+
Inference script for GIF-JAP-v0-2 (Gifu Wildlife Classifier - Central Japan)
|
| 3 |
+
|
| 4 |
+
This model classifies 13 species found in the Kuraiyama Experimental Forest (KEF) of
|
| 5 |
+
Gifu University. Trained on ~23,000 camera trap images to support efficient monitoring
|
| 6 |
+
of key wildlife species in central Japan (sika deer, wild boar, Asian black bear, Japanese serow).
|
| 7 |
+
|
| 8 |
+
Model: Gifu Wildlife v0.2
|
| 9 |
+
Input: 224x224 RGB images
|
| 10 |
+
Framework: PyTorch (ResNet50 with ImageNet initialization)
|
| 11 |
+
Classes: 13 Japanese species and taxonomic groups
|
| 12 |
+
Developer: Gifu University (Masaki Ando)
|
| 13 |
+
Citation: https://jglobal.jst.go.jp/en/detail?JGLOBAL_ID=201902236803626745
|
| 14 |
+
License: MIT
|
| 15 |
+
Info: https://github.com/gifu-wildlife/TrainingMdetClassifire
|
| 16 |
+
|
| 17 |
+
Note: Prototype model trained on limited and imbalanced data from KEF region.
|
| 18 |
+
|
| 19 |
+
Author: Peter van Lunteren
|
| 20 |
+
Created: 2026-01-14
|
| 21 |
"""
|
| 22 |
|
| 23 |
from __future__ import annotations
|