Initial RoadBench release: BEV manifests + FPV images + 8 task labels + train/test splits

README.md

@@ -47,6 +47,11 @@ FPV images are real in-vehicle camera photographs (anonymized) and are **include
 > construction, freshness) from the image used at dataset creation time. The geographic extent and
 > alignment are exact; only the pixel content can drift with time. This is an inherent property of
 > the coordinate-based distribution approach.
+>
+> **Reference imagery date.** The build-time BEV imagery (and the `rgb_histogram` fingerprints in
+> `bev_manifests/task1_2.jsonl`) correspond to **mid-2025 Google tiles** (acquired with the download
+> pipeline developed on 2025-07-19). Use [`verify_imagery.py`](verify_imagery.py) to measure how
+> much current tiles have drifted from this reference.
 
 ## Directory layout
 
@@ -156,6 +161,23 @@ The script writes `<id>.png` (Task 1/2/3) or `<image_id>.png` (Task 4/5) into `-
 **clean** satellite crop; the reference line shown to models is drawn separately from each label's
 `pixel_line` (in the reconstructed-image coordinate frame).
 
+### Checking reconstructed imagery
+
+Because Google updates its tiles over time, a reconstructed image may drift from the one used at
+dataset-build time. For Task 1/2, each `bev_manifests/task1_2.jsonl` record carries an `rgb_histogram`
+fingerprint of the build-time image; [`verify_imagery.py`](verify_imagery.py) compares your
+reconstructed images against it:
+
+```bash
+# after reconstructing Task 1/2 images into bev_images/task1_2/
+python verify_imagery.py verify --manifest bev_manifests/task1_2.jsonl --image-dir bev_images/task1_2
+```
+
+It prints cosine similarity stats (min/mean/median/max) and the 10 least-similar images. Rough guide:
+`>= 0.95` faithful · `0.80–0.95` minor drift (season / imagery refresh) · `< 0.80` substantial drift —
+re-check the crop window / coordinates. The fingerprint is resolution-independent (an L1-normalized
+binned RGB histogram), so reference and reconstructed images need not share pixel dimensions.
+
 ## Manifest schema
 
 JSON Lines, one record per BEV image. Only the fields needed to reconstruct the image are stored.
@@ -164,6 +186,9 @@ JSON Lines, one record per BEV image. Only the fields needed to reconstruct the
   (the WGS84 box to download), `crop_window` `{x0, y0, x1, y1}` (region to crop from the downloaded
   bbox tile, in bbox pixel coords).
 - **`task4_5.jsonl`**: `image_id`, `wgs84_bbox` (download the full bbox tile — no cropping).
+- **`rgb_histogram`** (Task 1/2 only): a compact fingerprint of the build-time image for drift
+  checking — base64 of little-endian uint16, 16 bins × 3 channels = 48 values, scaled by 65535
+  (see [Checking reconstructed imagery](#checking-reconstructed-imagery)).
 
 ## Label schema (`tasks/*/labels.jsonl`)
 
@@ -196,4 +221,4 @@ below (multiple letters = a lane allowing several directions):
 
 For example, `["B", "C", "CD"]` = a left-turn lane, a straight lane, and a straight+right-turn lane.
 `G` appears only in the FPV and cross-view designation tasks (5/7), matching the `variable` lane type
-used in the paper's MLLM prompt. `road_type` (Task 8) is `main` / `service`.
+used in the paper's MLLM prompt. `road_type` (Task 8) is `main` / `service`.

download_satellite.py

@@ -49,7 +49,7 @@ class BoundingBox:
 
 
 class TileCalculator:
-    """Web Mercator 瓦片计算（与原始 roadnetbenchmark.satellite 一致）。"""
+    """Web Mercator 瓦片计算（标准 Web Mercator 投影）。"""
     tile_size = 256
 
     @staticmethod
@@ -114,7 +114,7 @@ class GoogleEarthService:
 
 
 class Downloader:
-    """复刻原始 SatelliteImageDownloader 的下载/拼接/裁剪流程（max_tiles=160）。"""
+    """Google 卫星瓦片的下载/拼接/裁剪流程（max_tiles=160）。"""
 
     def __init__(self, proxy: Optional[str] = None, max_concurrent: int = 10,
                  cache_dir: str = "tile_cache", max_tiles: int = 160):