initialneil commited on
Commit
ef97c58
·
verified ·
1 Parent(s): c80409b

Add P2C1 metadata + SMPL-X + previews

Browse files
scripts/upload_capture.sh CHANGED
@@ -47,8 +47,11 @@ echo "=== [1/5] stage $CAPTURE ==============================================="
47
  --staging "$STAGING" $EXTRACT
48
 
49
  echo "=== [2/5] consolidate SMPL-X ==========================================="
 
 
 
50
  "$PY" "$HERE/consolidate_smplx.py" "$CAPTURE" --dataset-root "$DATASET_ROOT" \
51
- --out "$STAGING/data/$CAPTURE/smplx.npz"
52
 
53
  # capture.json is written before smplx.npz exists on a first run -> refresh it
54
  "$PY" "$HERE/stage_capture.py" "$CAPTURE" --dataset-root "$DATASET_ROOT" \
 
47
  --staging "$STAGING" $EXTRACT
48
 
49
  echo "=== [2/5] consolidate SMPL-X ==========================================="
50
+ # --expect-first-frame 0 is the d=0 guard: every capture's per-frame npz must start at
51
+ # frame 0. A source still carrying the old 1-based labelling is refused rather than
52
+ # silently published one frame out.
53
  "$PY" "$HERE/consolidate_smplx.py" "$CAPTURE" --dataset-root "$DATASET_ROOT" \
54
+ --out "$STAGING/data/$CAPTURE/smplx.npz" --expect-first-frame 0
55
 
56
  # capture.json is written before smplx.npz exists on a first run -> refresh it
57
  "$PY" "$HERE/stage_capture.py" "$CAPTURE" --dataset-root "$DATASET_ROOT" \
scripts/upload_remaining.sh ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env bash
2
+ # upload_remaining.sh [PxCy ...] -- upload a known-complete set of captures, SEQUENTIALLY.
3
+ #
4
+ # Tracking is finished, so there is no watching / polling / retry-until-ready logic here:
5
+ # this just walks a fixed list and uploads each one in turn.
6
+ #
7
+ # Sequential on purpose. The link to HF is ~1 MB/s and shared, so running captures in
8
+ # parallel does not add throughput, it just interleaves them and makes a stall in one
9
+ # capture look like a stall in all of them.
10
+ #
11
+ # Each capture is independent: a failure is logged and the loop moves on, so one bad
12
+ # capture cannot block the other five. upload_large_folder resumes per capture, so
13
+ # re-running this script after any failure costs only what is actually missing.
14
+ #
15
+ # Run detached (NOT tmux -- a tmux server restart on this shared box has killed an
16
+ # upload before):
17
+ # setsid nohup ./scripts/upload_remaining.sh > logs/seq.log 2>&1 < /dev/null &
18
+ set -uo pipefail
19
+
20
+ HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
21
+ BASE="$(dirname "$HERE")"
22
+ LOGDIR="$BASE/logs"; mkdir -p "$LOGDIR"
23
+
24
+ CAPTURES=("$@")
25
+ if [ ${#CAPTURES[@]} -eq 0 ]; then
26
+ CAPTURES=(P2C1 P2C2 P3C2 P4C1 P4C2 P5C2)
27
+ fi
28
+
29
+ STAMP="$(date +%Y%m%d_%H%M%S)"
30
+ SUMMARY="$LOGDIR/seq_summary_${STAMP}.txt"
31
+ echo "sequential upload of ${#CAPTURES[@]} captures: ${CAPTURES[*]}" | tee "$SUMMARY"
32
+ echo "started $(date)" | tee -a "$SUMMARY"
33
+
34
+ for c in "${CAPTURES[@]}"; do
35
+ log="$LOGDIR/seq_${c}_${STAMP}.log"
36
+ echo "" | tee -a "$SUMMARY"
37
+ echo "########## $c start $(date +%H:%M:%S) -> $log" | tee -a "$SUMMARY"
38
+ t0=$(date +%s)
39
+ "$HERE/upload_capture.sh" "$c" > "$log" 2>&1
40
+ rc=$?
41
+ t1=$(date +%s)
42
+ echo "########## $c exit=$rc elapsed=$(( (t1-t0)/60 ))m$(( (t1-t0)%60 ))s" | tee -a "$SUMMARY"
43
+ if [ $rc -ne 0 ]; then
44
+ echo " LAST 15 LINES OF $log:" | tee -a "$SUMMARY"
45
+ tail -15 "$log" | sed 's/^/ /' | tee -a "$SUMMARY"
46
+ fi
47
+ done
48
+
49
+ echo "" | tee -a "$SUMMARY"
50
+ echo "finished $(date)" | tee -a "$SUMMARY"
51
+ echo "ALL DONE" | tee -a "$SUMMARY"