ulmshare: align the data card with GitHub

ulmshare/README.md

@@ -7,7 +7,7 @@ task_categories:
 tags:
   - ultrasound
   - iq
-  - openh-rf
+  - open-h-rf
   - ulm
   - super-resolution
   - microbubbles
@@ -24,7 +24,7 @@ size_categories:
 
 <p align="center">
   <img width="500"
-    src="https://huggingface.co/datasets/nvidia/OpenH-RF/resolve/main/ulmshare/assets/hero.gif"
+    src="https://huggingface.co/datasets/nvidia/Open-H-RF/resolve/main/ulmshare/assets/hero.gif"
     alt="Ultrasound localization microscopy of the mouse brain." />
   <em style="color: #888;">
     ULM super-resolution reconstruction of a mouse brain.
@@ -79,7 +79,7 @@ Ultrasound localization microscopy: microbubble detection and localization, fram
 
 ## Processing the Dataset
 
-The acquisitions can be processed with the `reconstruct.py` [script](https://github.com/open-h/OpenH-RF/blob/main/datasets/ulmshare/reconstruct.py) as provided in the [OpenH-RF GitHub repository](https://github.com/open-h/OpenH-RF), together with the `pipeline_bmode.yaml` and `pipeline_tissue_suppression.yaml` definitions in this folder and the [zea library](https://github.com/tue-bmd/zea). The script streams the data from the Hugging Face Hub and writes a B-mode, a power-Doppler image and movie, and a ULM density map (the ULM steps live in [`ulm.py`](https://github.com/open-h/OpenH-RF/blob/main/datasets/ulmshare/ulm.py)).
+The acquisitions can be processed with the `reconstruct.py` [script](https://github.com/open-h/Open-H-RF/blob/main/datasets/ulmshare/reconstruct.py) as provided in the [Open-H-RF GitHub repository](https://github.com/open-h/Open-H-RF), together with the `pipeline_bmode.yaml` and `pipeline_tissue_suppression.yaml` definitions in this folder and the [zea library](https://github.com/tue-bmd/zea). The script streams the data from the Hugging Face Hub and writes a B-mode, a power-Doppler image and movie, and a ULM density map (the ULM steps live in [`ulm.py`](https://github.com/open-h/Open-H-RF/blob/main/datasets/ulmshare/ulm.py): per 400-frame buffer, buffers with tissue motion are skipped, microbubbles are localized against a per-depth noise floor, linked into tracks with a Hungarian tracker, and the tracks are drawn into a super-resolved density map).
 
 ## Dataset Format
 
@@ -137,7 +137,7 @@ Acquisition context with no zea spec field, stored as strings exactly as recorde
 
 ## Dataset Quantification
 
-**Current OpenH-RF release:** 103 HDF5 files; 20.74 TB (20,735,658,275,478 bytes) stored; root `zea_version` **0.1.4**. Sizes include all HDF5 contents and use decimal units (MB = 10^6 bytes, GB = 10^9 bytes, TB = 10^12 bytes), not decoded-array memory or original-source download sizes.
+**Current Open-H-RF release:** 103 HDF5 files; 20.74 TB (20,735,658,275,478 bytes) stored; root `zea_version` **0.1.4**. Sizes include all HDF5 contents and use decimal units (MB = 10^6 bytes, GB = 10^9 bytes, TB = 10^12 bytes), not decoded-array memory or original-source download sizes.
 
 - **Acquisitions:** 99, from 61 mice — one zea HDF5 file per acquisition.
 - **Stored HDF5 size:** 20.74 TB (20,735,658,275,478 bytes).