Upload folder using huggingface_hub

.ipynb_checkpoints/README-checkpoint.md

@@ -0,0 +1,137 @@
+# HIndoor-8K
+
+**HIndoor-8K** is the first metrically calibrated real-world benchmark of indoor
+RGB–D panoramas at native **8192×4096** (8K) resolution. It provides **49**
+equirectangular RGB panoramas, each paired with a **sparse metric depth map**
+rendered from a real LiDAR point cloud, across **5** representative indoor
+environments.
+
+> Released as a community resource for high-resolution 360° depth estimation.
+
+---
+
+## Contents
+
+```
+HIndoor-8K/
+├── README.md
+├── ich/          # corridor
+│   ├── rgb/      <timestamp>.jpg     RGB equirectangular panorama (8192×4096)
+│   └── depth/    <timestamp>.png     sparse metric depth (16-bit, millimetres)
+├── interrato/    # basement
+├── pt/           # project room
+├── sala/         # hall / meeting room
+└── unibs/        # laboratory
+```
+
+For every `rgb/<timestamp>.jpg` there is a matching `depth/<timestamp>.png`
+with the **same filename stem** (the capture timestamp).
+
+| Scene | Environment | # pairs |
+|-------|-------------|--------:|
+| ich       | corridor            | 7  |
+| interrato | basement            | 8  |
+| pt        | project room        | 10 |
+| sala      | hall / meeting room | 14 |
+| unibs     | laboratory          | 10 |
+| **Total** |                     | **49** |
+
+---
+
+## Data format
+
+**RGB** — `rgb/<timestamp>.jpg`
+- Real spherical photograph, equirectangular projection, **8192×4096**, 3-channel sRGB.
+- Real illumination, HDR/exposure variation, and sensor noise are preserved (not retouched).
+
+**Depth** — `depth/<timestamp>.png`
+- **16-bit single-channel PNG** (`I;16`, unsigned), **8192×4096**, pixel-aligned to the RGB.
+- Stored in **millimetres**: `depth_metres = png_value / 1000.0`.
+- **Sparse.** The map is a projection of a real LiDAR point cloud, so a large
+  fraction of pixels have **no valid return** (specular, distant, and
+  grazing-angle surfaces, plus the LiDAR's intrinsic angular sparsity).
+  **Invalid pixels are stored as `0`.** Typical valid coverage is ≈ 15–20 %.
+- **No densification / inpainting is applied** — this is the raw rendered depth.
+
+### Reading depth (Python)
+```python
+import numpy as np
+from PIL import Image
+Image.MAX_IMAGE_PIXELS = None
+
+d_mm  = np.asarray(Image.open("depth/<timestamp>.png"), dtype=np.float32)  # millimetres
+valid = d_mm > 0                       # validity mask (0 = no LiDAR return)
+d_m   = d_mm / 1000.0                  # metres
+```
+
+### Evaluation convention
+Compute all metrics **only over valid pixels** (`depth > 0`). In our paper we
+additionally cap the maximum depth at **16 m** for the reported HIndoor-8K
+numbers; adopt the same cap to reproduce them:
+```python
+mask = (d_m > 0) & np.isfinite(d_m) & (d_m <= 16.0)
+```
+Because the depth is **relative-free / metric but sparse**, predictions from
+relative-depth methods should be aligned to the ground truth (e.g. per-image
+median-ratio scale alignment) before computing error metrics.
+
+---
+
+## Acquisition
+
+Data is captured with a **professional indoor mobile-mapping platform** carrying
+rigidly coupled, mutually calibrated sensors:
+
+- **LiDAR:** dual Hesai XT32-class scanners (10–20 Hz), 32 channels,
+  31–40° vertical FoV, up to 0.1–0.2° angular resolution, 120 m (MT1) – 300 m (M2X)
+  range. Timestamped 3D points with reflectance / return-intensity.
+- **360° RGB camera:** a spherical camera rigidly mounted to the LiDAR rig,
+  capturing equirectangular panoramas at **8192×4096** (photo mode), with fixed
+  intrinsics and a known optical centre relative to the LiDAR.
+- **IMU:** 3-axis accelerometer + gyroscope for SLAM-based trajectory refinement.
+
+All sensors share a common reference frame. The pipeline outputs (i) a refined
+~10 Hz trajectory, (ii) synchronized LiDAR scans, and (iii) timestamped RGB panoramas.
+
+### Depth rendering
+Each RGB panorama's depth is produced by **off-screen spherical rendering** of the
+fused LiDAR point cloud, so the RGB stays a *real* photograph while depth comes
+from *real* geometry:
+
+1. **Pose interpolation** — the 10 Hz trajectory is interpolated to the exact RGB timestamp.
+2. **Spherical projection** — each 3D LiDAR point is projected to the spherical
+   camera (azimuth / elevation) and mapped to equirectangular coordinates.
+3. **Depth assignment** — each pixel stores the distance to the closest point
+   along its ray; occluded or inconsistent samples are discarded. Pixels that
+   receive no point remain `0` (→ the map is sparse).
+4. **Encoding** — depth is written as a 16-bit millimetre PNG at native resolution.
+
+---
+
+## Optional densification
+
+If a dense map is required, the sparse depth can be completed with an off-the-shelf
+depth-completion model (e.g. LDCM, *Large Depth Completion Model from Sparse
+Observations*, ICLR 2026). We deliberately ship the **raw sparse** depth so that
+users may choose their own completion (or none). All benchmark numbers in the
+paper are computed on the raw sparse depth with a validity mask.
+
+---
+
+## Citation
+
+If you use HIndoor-8K, please cite the FRED paper (bibentry to be added on release):
+
+```bibtex
+@article{fred_hindoor8k,
+  title  = {FRED: Full-Resolution Equirectangular Depth Estimation, and the HIndoor-8K Benchmark},
+  author = {Shah, U. and Pintore, G. and Tukur, M. and Zahoor, A. and Schneider, J. and
+            Sgrenzaroli, M. and Vassena, G. and V\'azquez, P.P. and Gobbetti, E. and Agus, M.},
+  journal= {Graphical Models},
+  year   = {2026}
+}
+```
+
+## License
+
+To be specified on public release. The dataset is provided for research use.

README.md

@@ -1,489 +1,137 @@
----
-task_categories:
-- depth-estimation
-language:
-- en
-size_categories:
-- n<1K
----
-
 # HIndoor-8K
 
-**HIndoor-8K** is a metrically calibrated, real-world benchmark for indoor panoramic depth estimation at native **8192 × 4096** resolution.
+**HIndoor-8K** is the first metrically calibrated real-world benchmark of indoor
+RGB–D panoramas at native **8192×4096** (8K) resolution. It provides **49**
+equirectangular RGB panoramas, each paired with a **sparse metric depth map**
+rendered from a real LiDAR point cloud, across **5** representative indoor
+environments.
 
-The dataset contains **49 equirectangular RGB panoramas**, each paired with a pixel-aligned **sparse metric depth map** rendered from a real LiDAR point cloud. The data covers **five indoor environments** and is intended primarily for evaluating high-resolution, full-resolution, and geometry-aware 360° depth-estimation methods.
-
-> HIndoor-8K is released as a community resource for high-resolution panoramic depth estimation.
+> Released as a community resource for high-resolution 360° depth estimation.
 
 ---
 
-## Highlights
-
-- Real indoor spherical RGB imagery.
-- Native **8K equirectangular resolution**.
-- Metric depth derived from real LiDAR measurements.
-- Pixel-aligned RGB and depth pairs.
-- Raw sparse depth without densification or inpainting.
-- Five representative indoor environments.
-- Evaluation protocol for metric and relative-depth models.
-
----
-
-## Dataset Overview
-
-| Property | Value |
-|---|---|
-| Number of RGB–depth pairs | 49 |
-| RGB resolution | 8192 × 4096 |
-| Depth resolution | 8192 × 4096 |
-| Projection | Equirectangular |
-| RGB format | JPEG |
-| Depth format | 16-bit PNG |
-| Depth unit | Millimetres |
-| Invalid depth value | 0 |
-| Typical valid coverage | Approximately 15–20% |
-| Number of environments | 5 |
-| Recommended evaluation cap | 16 metres |
+## Contents
 
----
-
-## Directory Structure
-
-```text
+```
 HIndoor-8K/
 ├── README.md
-├── ich/
-│   ├── rgb/
-│   │   └── <timestamp>.jpg
-│   └── depth/
-│       └── <timestamp>.png
-├── interrato/
-│   ├── rgb/
-│   └── depth/
-├── pt/
-│   ├── rgb/
-│   └── depth/
-├── sala/
-│   ├── rgb/
-│   └── depth/
-└── unibs/
-    ├── rgb/
-    └── depth/
-```
-
-For each RGB panorama:
-
-```text
-<scene>/rgb/<timestamp>.jpg
-```
-
-there is a matching depth map:
-
-```text
-<scene>/depth/<timestamp>.png
+├── ich/          # corridor
+│   ├── rgb/      <timestamp>.jpg     RGB equirectangular panorama (8192×4096)
+│   └── depth/    <timestamp>.png     sparse metric depth (16-bit, millimetres)
+├── interrato/    # basement
+├── pt/           # project room
+├── sala/         # hall / meeting room
+└── unibs/        # laboratory
 ```
 
-The RGB image and depth map use the same filename stem, corresponding to the capture timestamp.
-
----
-
-## Scene Distribution
+For every `rgb/<timestamp>.jpg` there is a matching `depth/<timestamp>.png`
+with the **same filename stem** (the capture timestamp).
 
-| Scene | Environment | Number of pairs |
-|---|---|---:|
-| `ich` | Corridor | 7 |
-| `interrato` | Basement | 8 |
-| `pt` | Project room | 10 |
-| `sala` | Hall or meeting room | 14 |
-| `unibs` | Laboratory | 10 |
-| **Total** |  | **49** |
+| Scene | Environment | # pairs |
+|-------|-------------|--------:|
+| ich       | corridor            | 7  |
+| interrato | basement            | 8  |
+| pt        | project room        | 10 |
+| sala      | hall / meeting room | 14 |
+| unibs     | laboratory          | 10 |
+| **Total** |                     | **49** |
 
 ---
 
-## Data Format
-
-### RGB Panoramas
-
-RGB files are stored under:
-
-```text
-<scene>/rgb/<timestamp>.jpg
-```
-
-Each RGB panorama has the following properties:
-
-- Resolution: **8192 × 4096**
-- Projection: equirectangular
-- Channels: three-channel sRGB
-- Source: real spherical photography
-- Format: JPEG
-- Real illumination, exposure variation, stitching artefacts, and sensor noise are preserved
-
-### Sparse Metric Depth
-
-Depth files are stored under:
-
-```text
-<scene>/depth/<timestamp>.png
-```
-
-Each depth map has the following properties:
-
-- Resolution: **8192 × 4096**
-- Format: unsigned 16-bit single-channel PNG
-- PIL mode: `I;16`
-- Unit: millimetres
-- Invalid value: `0`
-- Typical valid coverage: approximately 15–20%
-- No densification or inpainting is applied
+## Data format
 
-Convert stored depth values to metres using:
+**RGB** — `rgb/<timestamp>.jpg`
+- Real spherical photograph, equirectangular projection, **8192×4096**, 3-channel sRGB.
+- Real illumination, HDR/exposure variation, and sensor noise are preserved (not retouched).
 
-```python
-depth_m = depth_mm / 1000.0
-```
-
-The depth maps are sparse because they are generated by projecting real LiDAR observations into the panoramic camera view. Pixels may be invalid because of LiDAR angular sparsity, occlusion, distance, grazing-angle surfaces, reflective materials, or missing returns.
-
-A value of `0` means that no valid LiDAR observation is available. It does not represent a physical distance of zero metres.
-
----
-
-## Loading the Dataset
-
-### Load an RGB–Depth Pair
+**Depth** — `depth/<timestamp>.png`
+- **16-bit single-channel PNG** (`I;16`, unsigned), **8192×4096**, pixel-aligned to the RGB.
+- Stored in **millimetres**: `depth_metres = png_value / 1000.0`.
+- **Sparse.** The map is a projection of a real LiDAR point cloud, so a large
+  fraction of pixels have **no valid return** (specular, distant, and
+  grazing-angle surfaces, plus the LiDAR's intrinsic angular sparsity).
+  **Invalid pixels are stored as `0`.** Typical valid coverage is ≈ 15–20 %.
+- **No densification / inpainting is applied** — this is the raw rendered depth.
 
+### Reading depth (Python)
 ```python
-from pathlib import Path
-
 import numpy as np
 from PIL import Image
-
 Image.MAX_IMAGE_PIXELS = None
 
-root = Path("HIndoor-8K")
-scene = "ich"
-timestamp = "<timestamp>"
-
-rgb_path = root / scene / "rgb" / f"{timestamp}.jpg"
-depth_path = root / scene / "depth" / f"{timestamp}.png"
-
-rgb = Image.open(rgb_path).convert("RGB")
-
-depth_mm = np.asarray(
-    Image.open(depth_path),
-    dtype=np.float32,
-)
-
-valid_mask = depth_mm > 0
-depth_m = depth_mm / 1000.0
-
-print("RGB size:", rgb.size)
-print("Depth shape:", depth_m.shape)
-print("Valid coverage:", valid_mask.mean())
+d_mm  = np.asarray(Image.open("depth/<timestamp>.png"), dtype=np.float32)  # millimetres
+valid = d_mm > 0                       # validity mask (0 = no LiDAR return)
+d_m   = d_mm / 1000.0                  # metres
 ```
 
-### Enumerate All Samples
-
+### Evaluation convention
+Compute all metrics **only over valid pixels** (`depth > 0`). In our paper we
+additionally cap the maximum depth at **16 m** for the reported HIndoor-8K
+numbers; adopt the same cap to reproduce them:
 ```python
-from pathlib import Path
-
-root = Path("HIndoor-8K")
-scenes = ["ich", "interrato", "pt", "sala", "unibs"]
-
-samples = []
-
-for scene in scenes:
-    rgb_dir = root / scene / "rgb"
-    depth_dir = root / scene / "depth"
-
-    for rgb_path in sorted(rgb_dir.glob("*.jpg")):
-        depth_path = depth_dir / f"{rgb_path.stem}.png"
-
-        if not depth_path.exists():
-            raise FileNotFoundError(
-                f"Missing depth map for {rgb_path}"
-            )
-
-        samples.append(
-            {
-                "scene": scene,
-                "timestamp": rgb_path.stem,
-                "rgb_path": rgb_path,
-                "depth_path": depth_path,
-            }
-        )
-
-print(f"Found {len(samples)} RGB–depth pairs.")
+mask = (d_m > 0) & np.isfinite(d_m) & (d_m <= 16.0)
 ```
-
----
-
-## Evaluation Protocol
-
-All metrics must be computed only over valid depth pixels.
-
-### Validity Mask
-
-```python
-mask = (
-    (depth_m > 0)
-    & np.isfinite(depth_m)
-)
-```
-
-For the HIndoor-8K results reported in the FRED paper, the maximum ground-truth depth is capped at **16 metres**:
-
-```python
-mask = (
-    (depth_m > 0)
-    & np.isfinite(depth_m)
-    & (depth_m <= 16.0)
-)
-```
-
-Predictions must be finite and spatially aligned with the ground-truth depth map.
-
-### Metric-Depth Models
-
-Models that directly predict metric depth should normally be evaluated without per-image scale alignment.
-
-### Relative-Depth Models
-
-Relative-depth predictions should be aligned to the valid ground-truth observations before metric evaluation.
-
-A median-ratio alignment can be applied as follows:
-
-```python
-valid_prediction = (
-    mask
-    & np.isfinite(prediction)
-    & (prediction > 0)
-)
-
-scale = (
-    np.median(depth_m[valid_prediction])
-    / np.median(prediction[valid_prediction])
-)
-
-aligned_prediction = prediction * scale
-```
-
-The alignment method must be reported with the results.
-
-### Recommended Metrics
-
-Recommended evaluation metrics include:
-
-- Absolute Relative Error, `AbsRel`
-- Squared Relative Error, `SqRel`
-- Root Mean Squared Error, `RMSE`
-- Logarithmic RMSE
-- Scale-Invariant Log Error, `SILog`
-- Mean Absolute Error, `MAE`
-- Threshold accuracy, δ < 1.25
-- Threshold accuracy, δ < 1.25²
-- Threshold accuracy, δ < 1.25³
-
-Reported results should specify:
-
-- Maximum evaluation depth
-- Native or resized prediction resolution
-- Metric or relative-depth prediction
-- Scale-alignment method, if any
-- Interpolation method
-- Tiling or stitching strategy
-- Post-processing
-- Whether spherical-area weighting was used
+Because the depth is **relative-free / metric but sparse**, predictions from
+relative-depth methods should be aligned to the ground truth (e.g. per-image
+median-ratio scale alignment) before computing error metrics.
 
 ---
 
 ## Acquisition
 
-The dataset was captured using a professional indoor mobile-mapping platform carrying rigidly coupled and mutually calibrated sensors.
-
-### LiDAR
-
-The acquisition platform uses dual Hesai XT32-class LiDAR scanners with approximately:
-
-- 10–20 Hz scan frequency
-- 32 channels
-- 31–40° vertical field of view
-- Up to approximately 0.1–0.2° angular resolution
-- Nominal range between approximately 120 and 300 metres, depending on the sensor configuration
-- Timestamped three-dimensional points with reflectance or return intensity
-
-### Spherical RGB Camera
+Data is captured with a **professional indoor mobile-mapping platform** carrying
+rigidly coupled, mutually calibrated sensors:
 
-The spherical camera:
+- **LiDAR:** dual Hesai XT32-class scanners (10–20 Hz), 32 channels,
+  31–40° vertical FoV, up to 0.1–0.2° angular resolution, 120 m (MT1) – 300 m (M2X)
+  range. Timestamped 3D points with reflectance / return-intensity.
+- **360° RGB camera:** a spherical camera rigidly mounted to the LiDAR rig,
+  capturing equirectangular panoramas at **8192×4096** (photo mode), with fixed
+  intrinsics and a known optical centre relative to the LiDAR.
+- **IMU:** 3-axis accelerometer + gyroscope for SLAM-based trajectory refinement.
 
-- Is rigidly mounted relative to the LiDAR system
-- Captures equirectangular panoramas
-- Produces 8192 × 4096 images in photo mode
-- Uses fixed intrinsic calibration
-- Has a known optical-centre transformation relative to the LiDAR frame
-- Records timestamped images for trajectory synchronization
+All sensors share a common reference frame. The pipeline outputs (i) a refined
+~10 Hz trajectory, (ii) synchronized LiDAR scans, and (iii) timestamped RGB panoramas.
 
-### Inertial Measurement Unit
+### Depth rendering
+Each RGB panorama's depth is produced by **off-screen spherical rendering** of the
+fused LiDAR point cloud, so the RGB stays a *real* photograph while depth comes
+from *real* geometry:
 
-A three-axis accelerometer and gyroscope support SLAM-based trajectory estimation and refinement.
-
-The processing pipeline produces:
-
-1. A refined trajectory at approximately 10 Hz
-2. Synchronized LiDAR scans
-3. Timestamped RGB panoramas
-4. A fused three-dimensional point cloud
-5. Calibrated transformations between the sensors
+1. **Pose interpolation** — the 10 Hz trajectory is interpolated to the exact RGB timestamp.
+2. **Spherical projection** — each 3D LiDAR point is projected to the spherical
+   camera (azimuth / elevation) and mapped to equirectangular coordinates.
+3. **Depth assignment** — each pixel stores the distance to the closest point
+   along its ray; occluded or inconsistent samples are discarded. Pixels that
+   receive no point remain `0` (→ the map is sparse).
+4. **Encoding** — depth is written as a 16-bit millimetre PNG at native resolution.
 
 ---
 
-## Depth Rendering
-
-Each depth map is generated through off-screen spherical rendering of the fused LiDAR point cloud.
-
-The process consists of:
+## Optional densification
 
-1. **Pose interpolation**  
-   The reconstructed trajectory is interpolated to the exact RGB capture timestamp.
-
-2. **Coordinate transformation**  
-   LiDAR points are transformed into the spherical-camera coordinate frame.
-
-3. **Spherical projection**  
-   Each 3D point is converted to azimuth and elevation and mapped to an equirectangular pixel.
-
-4. **Depth assignment**  
-   Each pixel stores the distance to the closest consistent LiDAR point along the corresponding viewing direction.
-
-5. **Visibility filtering**  
-   Occluded, inconsistent, or invalid projected points are discarded.
-
-6. **Sparse-map preservation**  
-   Pixels receiving no valid point remain zero.
-
-7. **Encoding**  
-   Valid depth values are converted to millimetres and stored as 16-bit PNG files.
-
-The RGB images remain real photographs, while the depth maps originate from real geometric measurements.
-
----
-
-## Optional Depth Completion
-
-HIndoor-8K deliberately provides raw sparse depth.
-
-Researchers requiring dense depth may apply an external depth-completion method, such as LDCM or another sparse-to-dense model. Completed depth must be treated as model-generated data rather than original ground truth.
-
-Experiments using completed depth should report:
-
-- Completion architecture
-- Model checkpoint
-- Input preprocessing
-- Output resolution
-- Confidence filtering
-- Training data
-- Whether completion was used for training, evaluation, or both
-
-All benchmark results reported in the FRED paper are calculated using the original sparse depth and a valid-pixel mask.
-
----
-
-## Intended Uses
-
-HIndoor-8K is intended for research involving:
-
-- Monocular panoramic depth estimation
-- Full-resolution depth inference
-- Equirectangular image understanding
-- High-resolution depth estimation
-- Metric-depth evaluation
-- Relative-depth scale alignment
-- Sparse depth completion
-- Cross-domain depth evaluation
-- Panorama-seam analysis
-- Spherical geometry-aware neural networks
-- Tiled and patch-based inference
-
-The dataset is primarily an evaluation benchmark and is not large enough to serve as a standalone training dataset for modern deep networks.
-
----
-
-## Limitations
-
-Users should consider the following limitations:
-
-- The dataset contains only 49 samples.
-- It covers five indoor environments.
-- Ground-truth depth is sparse and non-uniformly distributed.
-- Metrics only evaluate locations containing valid LiDAR observations.
-- Reflective, transparent, dark, or distant surfaces may have missing returns.
-- Small calibration or synchronization errors may affect object boundaries.
-- The equirectangular projection introduces latitude-dependent distortion.
-- Native 8K processing requires substantial memory and computation.
-- Per-image scale alignment can significantly affect relative-depth results.
-- Repeated model tuning on the complete benchmark may cause overfitting.
-- Results should not be interpreted as representative of all indoor environments.
-
----
-
-## Recommended Reporting Practice
-
-To support reproducibility, report:
-
-- Dataset version
-- Evaluated scenes and images
-- Prediction resolution
-- Maximum depth
-- Validity mask
-- Scale alignment
-- Interpolation method
-- Tiling strategy
-- Seam handling
-- Circular padding
-- Post-processing
-- Overall metrics
-- Per-scene metrics
-- Runtime
-- Peak GPU memory
-- Hardware configuration
+If a dense map is required, the sparse depth can be completed with an off-the-shelf
+depth-completion model (e.g. LDCM, *Large Depth Completion Model from Sparse
+Observations*, ICLR 2026). We deliberately ship the **raw sparse** depth so that
+users may choose their own completion (or none). All benchmark numbers in the
+paper are computed on the raw sparse depth with a validity mask.
 
 ---
 
 ## Citation
 
-Please cite the FRED paper when using HIndoor-8K:
+If you use HIndoor-8K, please cite the FRED paper (bibentry to be added on release):
 
 ```bibtex
 @article{fred_hindoor8k,
-  title   = {FRED: Full-Resolution Equirectangular Depth Estimation, and the HIndoor-8K Benchmark},
-  author  = {Shah, U. and Pintore, G. and Tukur, M. and Zahoor, A. and Schneider, J. and
-             Sgrenzaroli, M. and Vassena, G. and V{\'a}zquez, P. P. and Gobbetti, E. and Agus, M.},
-  journal = {Graphical Models},
-  year    = {2026}
+  title  = {FRED: Full-Resolution Equirectangular Depth Estimation, and the HIndoor-8K Benchmark},
+  author = {Shah, U. and Pintore, G. and Tukur, M. and Zahoor, A. and Schneider, J. and
+            Sgrenzaroli, M. and Vassena, G. and V\'azquez, P.P. and Gobbetti, E. and Agus, M.},
+  journal= {Graphical Models},
+  year   = {2026}
 }
 ```
 
----
-
 ## License
 
-**License:** Creative Commons Attribution-NonCommercial 4.0
-
----
-
-## Dataset Version
-
-- **Version:** `1.0.0`
-- **Release date:** 02/08/2026
-- **Last updated:** 02/08/2026
-
----
-
-## Contact
-
-For questions about the dataset, evaluation protocol, or licensing:
 
-- **Name:** Uzair Shah
-- **Email:** ushah@hbku.edu.qa, magus@hbku.edu.qa
-[More Information Needed]