Hv add wac

algorithm_catalog/wur/worldagrocommodities/benchmark_scenarios/wac_inference.json

@@ -0,0 +1,32 @@
+[
+    {
+        "id": "WorldAgriCommodities",
+        "type": "openeo",
+        "description": "openEO pipeline for agricultural feature extraction and model inference using Sentinel-1 and Sentinel-2 data of the WorldAgriCommodities project",
+        "backend": "openeo.dataspace.copernicus.eu",
+        "process_graph": {
+            "parcel_delineation1": {
+                "process_id": "wac_inference",
+                "namespace": "https://raw.githubusercontent.com/ESA-APEx/apex_algorithms/96dfa13ba74ccbfb3432c6ebd38c42e704526e79/algorithm_catalog/wur/worldagrocommodities/records/wac_inference.json",
+                "arguments": {
+                    "spatial_extent": {
+			            "west": 716200,
+                        "south": 605530,
+                        "east": 722380,
+                        "north": 612410,
+                        "crs": "EPSG:32630"
+		            },
+                    "temporal_extent": [
+                        "2023-06-01",
+                        "2023-09-01"
+                    ]
+                },
+                "result": true
+            }
+        },
+        "reference_data": {
+	   }
+    }
+]
+
+

algorithm_catalog/wur/worldagrocommodities/openeo_udp/README.md

@@ -0,0 +1,38 @@
+# World Agri Commodities (WAC) Processing Pipeline
+This repository implements an end-to-end openEO pipeline for agricultural feature extraction and model inference using Sentinel-1 and Sentinel-2 data, a suite of preprocessing steps, vegetation indices, and a sliding-window ONNX model. The ML model was trained to classify agriculture activity for specific commodities in the context of ESA's World Agro Commodities project
+
+## Algorithm and Workflow
+### 1 Data Loading
+
+Sentinel-2 L2A (optical) and Scene Classification Layer (SCL) for cloud masking
+
+Sentinel-1 global mosaics (VV and VH)
+
+Digital Elevation Model (COPERNICUS_30)
+
+### 2 Preprocessing
+
+Spatial reprojection & resampling (CRS EPSG:3035, resolution 10 m)
+
+Cloud mask dilation on SCL → mask optical bands
+
+Monthly aggregation (mean) on Sentinel-2, decimation on Sentinel-1 (log transform), DEM temporal mean
+
+Compute vegetation indices (NDVI, NDRE, EVI)
+
+Apply lat/lon UDF to add geographic coordinates
+
+### 3 Normalization
+
+Optical bands: scale, log-transform, nonlinear sigmoid normalization
+
+Linear bands: min–max clipping and scaling for indices, radar backscatter, DEM, lat/lon
+
+### 4 Model Inference
+
+Patch-based sliding window (size 128×128 px, overlap 64 px) via apply_neighborhood UDF
+
+ONNX U-Net model loaded at runtime
+
+Output bands renamed to sigmoid_prob_class_0…prob_class_{N-1} plus prob_class
+