Update README.md

README.md

@@ -133,17 +133,17 @@ Contains cleaned image metadata and a rich `annotations` array generated by the
 ## Dataset Creation
 
 ### Data Collection and Processing
-[cite_start]The images were collected in **Meru County, Kenya** (spelled M-E-R-U) during the 2016 and 2018 census rallies[cite: 35, 43]. This dataset is the direct output of the GGR Pipeline, a 9-stage workflow designed for robust ecological tracking:
+The images were collected in **Meru County, Kenya** during the 2016 and 2018 census rallies. This dataset is the direct output of the GGR Pipeline, a 9-stage workflow designed for robust ecological tracking:
 
 1. **Import:** Parses raw field data, ingests card/camera metadata, and generates standard metadata descriptions, mapping absolute paths to the images.
 2. **Detection:** Utilizes YOLO to localize animals in the images, creating initial bounding boxes (`bbox`).
-3. [cite_start]**Species Classification:** Processes detections through BioCLIP to classify the species (e.g., Grevy's zebra vs. Plains zebra)[cite: 20].
+3. **Species Classification:** Processes detections through BioCLIP to classify the species (e.g., Grevy's zebra vs. Plains zebra).
 4. **Viewpoint Classification:** Determines the orientation of the detected animal (`up`, `front`, `back`, `left`, `right`).
-5. **Identifiable Annotation (IA) Classification:** Evaluates photographic quality and pose. [cite_start]Assigns a `CA_score` and a boolean flag indicating if the animal's features (e.g., stripes) are clear enough for individual identification[cite: 38].
-6. **IA Filtering:** Discards annotations below the identifiability threshold and simplifies viewpoints strictly to `left` or `right` flanks for Re-ID purposes.
-7. **Miew-Id (M-I-E-W-I-D):** Generates deep metric embeddings for all remaining high-quality annotations leveraging the Miew-Id feature extraction model.
-8. **Local Clusters and Alternatives (LCA) Algorithm:** Clusters the generated embeddings based on feature similarity, grouping images of the same individual animal together.
-9. **Post-processing and ID Assignment:** Applies biological consistency checks, resolves cluster overlaps, allows for human-in-the-loop manual verification, assigns the final `individual_id`, and integrates non-identifiable tracking links where necessary.
+5. **Identifiable Annotation (IA) Classification:** Evaluates photographic quality and pose. Assigns a `CA_score` and a boolean flag indicating if the animal's features (e.g., stripes) are clear enough for individual identification.
+7. **IA Filtering:** Discards annotations below the identifiability threshold and simplifies viewpoints strictly to `left` or `right` flanks for Re-ID purposes.
+8. **Miew-ID:** Generates deep metric embeddings for all remaining high-quality annotations leveraging the Miew-Id feature extraction model.
+9. **Local Clusters and Alternatives (LCA) Algorithm:** Clusters the generated embeddings based on feature similarity, grouping images of the same individual animal together.
+10. **Post-processing and ID Assignment:** Applies biological consistency checks, resolves cluster overlaps, allows for human-in-the-loop manual verification, assigns the final `individual_id`, and integrates non-identifiable tracking links where necessary.
 
 ---
 
@@ -151,8 +151,8 @@ Contains cleaned image metadata and a rich `annotations` array generated by the
 
 - **Missing Location Data:** When GPS coordinates could not be resolved from field equipment or manual logs, fields for `latitude` and `longitude` (`gps_lat`/`gps_lon`) default to `-1.0`. Algorithms utilizing geometric filtering should catch these sentinel values.
 - **Viewpoint Filtering:** Because animal Re-ID relies heavily on flank stripe patterns, the pipeline aggressively filters for `left` and `right` viewpoints. Forward or backward-facing animals are detected but may be excluded from Re-ID tracking.
-- [cite_start]**Geographic Focus:** The spatial data anchors tightly to the **Meru County** ecosystem and adjacent northern rangelands in Kenya[cite: 10, 45]. Models trained heavily on this dataset may experience domain shift when deployed in different biomes or under disparate lighting conditions.
-- [cite_start]**Class Imbalance:** Due to the targeted focus of the Great Grevy's Rally, equid and giraffe classes heavily outnumber categories like `lion` or `rhino_black`[cite: 35, 37].
+- **Geographic Focus:** The spatial data anchors tightly to the **Meru County** ecosystem and adjacent northern rangelands in Kenya. Models trained heavily on this dataset may experience domain shift when deployed in different biomes or under disparate lighting conditions.
+- **Class Imbalance:** Due to the targeted focus of the Great Grevy's Rally, equid and giraffe classes heavily outnumber categories like `lion` or `rhino_black`.
 
 ---
 
@@ -176,8 +176,6 @@ If you use this dataset, please cite both the dataset release and the foundation
   address   = {Stanford, CA},
   year      = {2017}
 }
-
-
 ```
 
 ## Dataset Card Authors