Upload 3 files (#2)

- Upload 3 files (284563f7ac0103fdc3ae7265fb47caef58aa15ba)

README.md

@@ -25,13 +25,13 @@ metrics:
   - recall
   - f1
 
-base_model: "Roboflow/rf-detr-small"
+base_model: "Roboflow/rf-detr-nano"
 ---
 
 
-# RF-DETR Small Finetuned on Global Wheat Head Dataset
+# RF-DETR Nano Finetuned on Global Wheat Head Dataset
 
-Fine-tuned RF-DETR Small object detector on the **Global Wheat Head Dataset** benchmark dataset, trained and evaluated as part of [DetectionBench](https://github.com/dronefreak/DetectionBench) -- a framework for reproducibly benchmarking modern object detectors with identical training recipes and evaluation metrics across multiple real-world datasets.
+Fine-tuned RF-DETR Nano object detector on the **Global Wheat Head Dataset** benchmark dataset, trained and evaluated as part of [DetectionBench](https://github.com/dronefreak/DetectionBench) -- a framework for reproducibly benchmarking modern object detectors with identical training recipes and evaluation metrics across multiple real-world datasets.
 
 <br>
 
@@ -39,14 +39,14 @@ Fine-tuned RF-DETR Small object detector on the **Global Wheat Head Dataset** be
 <div style="display: flex; justify-content: center; align-items: center; gap: 8px; margin-bottom: 8px; flex-wrap: wrap;">
   <img src="https://img.shields.io/badge/Task-Object_Detection-blue?style=flat-square" alt="Task">
   <img src="https://img.shields.io/badge/Framework-RF--DETR-0aa1a7?style=flat-square" alt="Framework">
-  <img src="https://img.shields.io/badge/Base_Model-RF--DETR_Small-purple?style=flat-square" alt="Base Model">
+  <img src="https://img.shields.io/badge/Base_Model-RF--DETR_Nano-purple?style=flat-square" alt="Base Model">
 </div>
 
 <!-- ROW 2: Performance Metrics -->
 <div style="display: flex; justify-content: center; align-items: center; gap: 8px; margin-bottom: 8px; flex-wrap: wrap;">
-  <img src="https://img.shields.io/badge/mAP@50-64.51%25-success?style=flat-square" alt="mAP@50">
-  <img src="https://img.shields.io/badge/mAP@50:95-26.12%25-orange?style=flat-square" alt="mAP@50:95">
-  <img src="https://img.shields.io/badge/Params-32.1M-lightgrey?style=flat-square" alt="Params">
+  <img src="https://img.shields.io/badge/mAP@50-53.82%25-success?style=flat-square" alt="mAP@50">
+  <img src="https://img.shields.io/badge/mAP@50:95-19.65%25-orange?style=flat-square" alt="mAP@50:95">
+  <img src="https://img.shields.io/badge/Params-30.5M-lightgrey?style=flat-square" alt="Params">
 </div>
 
 <!-- ROW 3: Metadata -->
@@ -60,7 +60,7 @@ Fine-tuned RF-DETR Small object detector on the **Global Wheat Head Dataset** be
 ## Detection Showcase
 
 <p align="center">
-  <img src="gwhd_rfdetr-small_showcase.jpg" alt="Global Wheat Head Dataset Detection Demo" width="900">
+  <img src="gwhd_rfdetr-nano_showcase.jpg" alt="Global Wheat Head Dataset Detection Demo" width="900">
 </p>
 
 ---
@@ -69,12 +69,12 @@ Fine-tuned RF-DETR Small object detector on the **Global Wheat Head Dataset** be
 
 | Metric     | Score (%)       |
 | ---------- | --------------- |
-| mAP@50     | 64.51     |
-| mAP@50-95  | 26.12   |
-| Precision  | 77.6 |
-| Recall     | 63.61    |
-| F1 Score   | 69.91        |
-| Parameters | 32.1M    |
+| mAP@50     | 53.82     |
+| mAP@50-95  | 19.65   |
+| Precision  | 72.52 |
+| Recall     | 53.64    |
+| F1 Score   | 61.67        |
+| Parameters | 30.5M    |
 | FLOPs      | N/A (not published upstream)     |
 
 ---
@@ -106,7 +106,25 @@ Every model DetectionBench has trained and evaluated on Global Wheat Head Datase
 
 | Class                      | mAP@50          | mAP@50-95         |
 | -------------------------- | --------------- | ----------------- |
-| wheat_head       | 64.51 | 26.12 |
+| wheat_head       | 53.82 | 19.65 |
+---
+
+## Per-Country Performance
+
+Domain shift can matter more than the aggregate score above for field deployment, so this evaluates the same test split broken down by the contributing country/institution, using per-image domain metadata (`domain_metadata.json` in this repository) compiled by this project for this stratified evaluation -- not a file shipped with the original GWHD release. Each row below is computed by re-running this exact model's evaluation restricted to that country's images only -- the same mAP definition as the aggregate number above (Ultralytics' `model.val()` for YOLO, Supervision's `MeanAveragePrecision` for RF-DETR), just on a filtered subset, not a separate metric implementation. One test image with no resolvable country in the source metadata (a documented upstream duplicate-filename quirk) is excluded from every row below.
+
+| Country                    | mAP@50          | mAP@50-95         | Test Images        |
+| --------------------------- | --------------- | ----------------- | ------------------ |
+| Australia       | 35.4 | 10.86 | 281 |
+| China       | 79.78 | 33.58 | 200 |
+| Japan       | 61.07 | 30.42 | 60 |
+| Mexico       | 54.51 | 19.27 | 205 |
+| Sudan       | 61.42 | 24.15 | 30 |
+| US       | 58.7 | 20.9 | 605 |
+
+
+See `country_breakdown.json` (results) and `domain_metadata.json` (the country/growth-stage mapping used to compute them) in this repository for the raw data behind this table.
+
 ---
 
 ## Evaluation Visualizations
@@ -141,11 +159,11 @@ from huggingface_hub import hf_hub_download
 import rfdetr
 
 weights = hf_hub_download(
-    repo_id="dronefreak/gwhd-rfdetr-small",
+    repo_id="dronefreak/gwhd-rfdetr-nano",
     filename="checkpoint_best_total.pth"
 )
 
-model = rfdetr.RFDETRSmall(pretrain_weights=weights)
+model = rfdetr.RFDETRNano(pretrain_weights=weights)
 ```
 
 ### Run Inference
@@ -163,10 +181,10 @@ detections = model.predict("image.jpg", threshold=0.25)
 | Framework        | RF-DETR                  |
 | Training Toolkit | DetectionBench                   |
 | Epochs (configured max) | 500 |
-| Epochs (actually trained) | 124 |
+| Epochs (actually trained) | 110 |
 | Early Stopping Patience | 100 |
-| Batch Size | 6 |
-| Resolution | 512 |
+| Batch Size | 7 |
+| Resolution | 384 |
 | Optimizer | adamw |
 | Learning Rate | 0.0001 |
 | Seed | 42 |
@@ -178,7 +196,9 @@ detections = model.predict("image.jpg", threshold=0.25)
 checkpoint_best_total.pth
 metrics.csv
 config.json
-gwhd_rfdetr-small_showcase.jpg
+country_breakdown.json
+domain_metadata.json
+gwhd_rfdetr-nano_showcase.jpg
 README.md
 ```
 
@@ -221,7 +241,7 @@ If you find this model useful, please consider starring the repository.
 If you use this model in your research, please consider citing:
 
 1. The Global Wheat Head Dataset dataset (see below)
-2. The original RF-DETR Small architecture (see below)
+2. The original RF-DETR Nano architecture (see below)
 3. The other model architectures shown in the Model Zoo/External Comparison tables above, if you reference their results
 4. DetectionBench, the training/evaluation framework used to produce this checkpoint
 ```

country_breakdown.json

@@ -0,0 +1,32 @@
+{
+  "Australia": {
+    "mAP50": 0.3540254533290863,
+    "mAP50_95": 0.10858235508203506,
+    "num_images": 281
+  },
+  "China": {
+    "mAP50": 0.7978321313858032,
+    "mAP50_95": 0.3357720375061035,
+    "num_images": 200
+  },
+  "Japan": {
+    "mAP50": 0.6107019186019897,
+    "mAP50_95": 0.30420443415641785,
+    "num_images": 60
+  },
+  "Mexico": {
+    "mAP50": 0.5450758934020996,
+    "mAP50_95": 0.1927119940519333,
+    "num_images": 205
+  },
+  "Sudan": {
+    "mAP50": 0.6141822934150696,
+    "mAP50_95": 0.24153614044189453,
+    "num_images": 30
+  },
+  "US": {
+    "mAP50": 0.5870068073272705,
+    "mAP50_95": 0.2090480625629425,
+    "num_images": 605
+  }
+}