PKG-9 · project Kirioll · 2026-07-07
Detector landscape 2026 · small-object strategies · prior art
→ / space / swipe to advance
Real raster/label format still open (src/dataset/rasters.py is a stub); only the synthetic 256×256 path runs today. Note: the rendered sections are sparse stroke drawings — thin point traces on empty background, not dense rasters — a real domain gap vs natural-image pretraining, favoring foundation-model backbones (DINOv2 in RF-DETR).
Full-image detection is a tiny-object regime; a native-resolution crop is not. The scale problem is a tiling problem first, a loss problem second.
Numbers from the NWD paper (arXiv:2110.13389). IoU-thresholded anchor matching flips tiny objects between positive and negative on 1 px noise — assignment, not regression, starves them.
NWD (Wasserstein on Gaussian boxes, +6.7 AP on AI-TOD) and RFLA (receptive-field ranking, +4.0 AP) replace IoU in matching.
SAHI sliced fine-tuning + inference: +12–14 AP. Our 256 px crop sampling is exactly this.
Wise-IoU, Shape-IoU, Inner-IoU, Powerful-IoU: gradient at zero overlap, size-normalized penalties. Small gains next to 1–2. Focal/Varifocal for imbalance.
Copy-paste + oversample: +9.7% small-object AP (Kisantal 2019). Easy here — profiles are localized, rare types need it anyway.
Segmentation sidesteps all four — no anchors, no IoU matching. Residual thin-mask fragility → loss menu: clDice (topology), Focal Tversky (small-structure imbalance, ISBI 2019), Generalized Dice (inverse-squared-volume weights for rare type classes, 2017) on top of dice_focal.
Smallest variant of each; vendor-reported. Larger RF-DETRs reach 60.1 mAP; DEIMv2/D-FINE (Apache 2.0) sit near RF-DETR. YOLO26's STAL is the only YOLO mechanism aimed at small targets — it guarantees ≥4 anchor assignments for objects <8 px, the anchor-starvation fix productized. Practical gotcha: RF-DETR inputs start at 384 px (must divide by patch×windows; 384/512/576/704) — 256 px crops need padding or a 384 px export. Every modern mainstream YOLO is AGPL unless Enterprise-licensed (older Apache-2.0 YOLOs exist: YOLOX 2021, PP-YOLOE, DAMO-YOLO); RF-DETR also has the best domain-transfer record (RF100-VL 85–89 AP50).
YOLO on 1 cm LiDAR cross-section images, 13 iRAP classes incl. metal + concrete barriers — concrete AP 0.98 (Sensors 22:5510). Validates our exact modality.
Per-section CNNs on point-cloud + image slices, fused for guardrail extraction across diverse designs (IJAT 15:258).
RGB-D segmentation + metric dimensions → type from a European guardrail catalog, trained purely on synthetic data (ICMVA). Nothing published at EDSP/ESP/Super-Rail granularity.
Vertical-profile/DoN tracking (Hou 2022), filtering + DBSCAN (Gao 2022), profile template matching for rails (Arastounia 2016) — classical fallback for DXF-profile type matching.
| # | Experiment | Notes | |
|---|---|---|---|
| 1 | Segmentation baseline (U-Net, dice_focal, 256 px crops) | Current plan; add clDice if masks are thin strips | keep |
| 2 | RF-DETR Nano/Small on crops | Apache 2.0; RF-DETR-Seg gives per-instance masks → feeds type classification. Needs 384 px inputs (256 unsupported) → export 384 px crops or pad | clean license |
| 3 | YOLO26 on crops | STAL + NMS-free; experiment-only unless Enterprise | AGPL |
| 4 | Full-section detection | Only with SAHI tiling (reference recipe: 256 px slices, 20% overlap, cross-slice NMS merge) + NWD/RFLA assignment + Wise-IoU + P2 head | last resort |
| 5 | Type classification per instance | DXF 14-point profiles → template-match baseline + synthetic pretraining à la Beleznai; copy-paste rare types | the open problem |
NWD 2110.13389 · RFLA 2208.08738 · SAHI 2202.06934 · small-object survey (TPAMI) 2207.14096 · Wise-IoU 2301.10051 · Kisantal 1902.07296 · clDice 2003.07311 · Focal Tversky 1810.07842 · Generalized Dice 1707.03237 · RF-DETR 2511.09554 · RF-DETR training docs · YOLOv12 2502.12524 · YOLOv13 2506.17733 · YOLO26 docs · DEIMv2 2509.20787 · Brkić 10.3390/s22155510 · Matsumoto 10.20965/ijat.2021.p0258 · Beleznai 10.1145/3589572.3589597
Full annotated list: vault note 2026-07-07 Guardrail detection from highway cross-sections (Project notes / Point cloud analysis on linear structures).