LunaDream74 commited on
Commit
93a02e3
·
verified ·
1 Parent(s): 6060a5e

docs: dataset card with nuScenes attribution

Browse files
Files changed (1) hide show
  1. README.md +17 -5
README.md CHANGED
@@ -37,8 +37,18 @@ thumbs.tar -> repo root <- fetch this one
37
  |---|---|---|
38
  | `artifacts/vectors.npy` | 21 MB | 6858 × 768 float32, L2-normalised SigLIP2 image embeddings |
39
  | `artifacts/caption_vectors.npy` | 10.5 MB | 6858 × 384 float32, MiniLM embeddings of generated captions |
40
- | `thumbs.tar` | 82 MB | all 6,858 thumbnails, flat and uncompressed, faces/plates redacted |
41
- | `*.jpg` | 70 MB | the same thumbnails loose, 320×180, downscaled from 1600×900 |
 
 
 
 
 
 
 
 
 
 
42
 
43
  **The thumbnails are redacted.** Faces and licence plates are mosaicked at source resolution before
44
  the downscale — 5,511 regions across the corpus, 2,616 frames (38.1%) carrying at least one.
@@ -64,9 +74,11 @@ unauthenticated caller for. It returned 429 at roughly the 490th file on a CI ru
64
  minutes in, leaving a half-populated directory. One request cannot half-arrive.
65
 
66
  The tarball is uncompressed deliberately: measured over 200 of these thumbnails, gzip -6 saves 2.1%,
67
- because JPEG is already DCT-compressed. It is larger than the sum of its contents (82 MB against 70)
68
- because tar pads every member to a 512-byte boundary and these files are small — that padding is the
69
- price of the single request, and it compresses away in transit.
 
 
70
 
71
  Row order in both matrices matches `frames.jsonl` in the application repo exactly, and the
72
  application re-checks that on load — a misaligned row would mean confident answers about the wrong
 
37
  |---|---|---|
38
  | `artifacts/vectors.npy` | 21 MB | 6858 × 768 float32, L2-normalised SigLIP2 image embeddings |
39
  | `artifacts/caption_vectors.npy` | 10.5 MB | 6858 × 384 float32, MiniLM embeddings of generated captions |
40
+ | `thumbs.tar` | 267 MB | all 6,858 thumbnails, flat and uncompressed, faces/plates redacted |
41
+ | `*.jpg` | 261 MB | the same thumbnails loose, 640×360, downscaled from 1600×900 |
42
+
43
+ **640×360 as of 2026-08-06, regenerated from the nuScenes source.** Thumbnail width is one constant,
44
+ `scripts/anonymise.py: THUMB_WIDTH`, raised off 320 on 2026-08-05; this published set predated that
45
+ change by a day and has now caught up with it. 320 was inherited from nuScenes and discards most of
46
+ the frame on modern footage, which is a 2× rather than 5× downscale.
47
+
48
+ The rewrite re-read the source images and re-applied redaction at source resolution, so **the privacy
49
+ figures below are reproduced exactly rather than carried over** — same 5,511 regions, same 2,616
50
+ frames, same 239/348/4,924 split, `privacy_unverified` still 0. Redaction geometry does not depend on
51
+ the output width, which is the point of mosaicking before the downscale.
52
 
53
  **The thumbnails are redacted.** Faces and licence plates are mosaicked at source resolution before
54
  the downscale — 5,511 regions across the corpus, 2,616 frames (38.1%) carrying at least one.
 
74
  minutes in, leaving a half-populated directory. One request cannot half-arrive.
75
 
76
  The tarball is uncompressed deliberately: measured over 200 of these thumbnails, gzip -6 saves 2.1%,
77
+ because JPEG is already DCT-compressed. It is larger than the sum of its contents (267 MB against
78
+ 261) because tar pads every member to a 512-byte boundary. **That overhead used to be the notable
79
+ part of this paragraph and no longer is** — at 320×180 the members averaged ~10 KB and padding cost
80
+ 17%; at 640×360 they average 37 KB and it costs 2%. Padding is a fixed ~0.75 KB per member, so it
81
+ matters in inverse proportion to file size.
82
 
83
  Row order in both matrices matches `frames.jsonl` in the application repo exactly, and the
84
  application re-checks that on load — a misaligned row would mean confident answers about the wrong