Viabig logoViabig
Services ↓
Bounding BoxPolygon & InstanceSemantic SegmentationKeypoint & LandmarkLiDAR / 3DNLP / Text LabelingAudio TranscriptionDataset CurationModel Training & Fine-TuningLLMs & RAG SystemsGenerative AIComputer VisionMLOps & InfraSports AnalyticsHealthcare & Life SciencesAutomotive & MobilityRetail & E-commerceManufacturing & Industry 4.0Financial ServicesLogistics & Supply ChainEnergy & Utilities
AboutCareersBlogContactBook a call →
Home/Services/Discipline 01
// Discipline 01

Production data, annotated by engineers.

Eight annotation modalities under one QA discipline. Two-pass review on every object, drift tracking by class, edge-case mining built into the workflow.

// annotation · live · 3 classesLIVEpedestrian · 0.97car · 0.94sign · 0.88annotated: 3 / 3 · QA: pass · iou_mean: 0.93frame_042
/ 1.1

Bounding Box

2D rectangles around objects of interest — the workhorse of detection training. Sounds simple. Done correctly at volume, it is not.

// bbox · 3 classes · 0.97 confpedestrian · 0.97car · 0.94sign · 0.88

// Details

  • Object detection (YOLO, DETR, RT-DETR)
  • Surveillance, retail analytics, autonomous driving
  • Two-pass QA with tightness IoU sampling against gold set
  • Class-balance reports per delivery

// Output formats

COCOYOLO TXTPascal VOCCSV
/ 1.2

Polygon & Instance Segmentation

Tight, class-aware masks for irregular shapes. Per-instance even when objects of the same class overlap.

// polygon · instanceperson · #1vehicle · #1

// Details

  • Instance segmentation (Mask R-CNN, Mask2Former)
  • Medical imaging, agriculture, defect inspection
  • Boundary precision ≤ 2px on long edges
  • Inter-annotator IoU ≥ 0.85 on convergence

// Output formats

COCO RLECOCO polygonPNG mask
/ 1.3

Semantic Segmentation

Per-pixel class maps. Dense, label-everything output for scene understanding tasks.

// semantic · 7 classesskyvegetationroadbuilding

// Details

  • Autonomous driving (Cityscapes-style labels)
  • Satellite / aerial imagery
  • Medical segmentation (organs, lesions)
  • CVAT with SAM-assisted brush, pixel-level QA overlays

// Output formats

Indexed PNGClass mapADE20K
/ 1.4

Keypoint & Landmark

Pose estimation, facial landmarks, skeletal annotation, and per-class custom keypoint schemes.

// keypoint · 17 pts

// Details

  • Pose estimation (sports, fitness, rehab)
  • Facial landmarks for AR / animation
  • Body 17 / 25 keypoint conventions
  • Facial 68 / 98 / 106 landmark schemes

// Output formats

COCO KeypointsMPIICustom JSON
/ 1.5

LiDAR / 3D Annotation

Point-cloud labeling, 3D cuboid boxes, and sensor-fusion datasets for AV and robotics.

// LiDAR · 3D cuboidvehicle · L 4.8m · W 1.9m · H 1.6m

// Details

  • Autonomous driving (LiDAR-only or fusion)
  • Robotics, drone perception
  • 3D cuboids with heading & velocity
  • Multi-sweep tracking IDs, image-LiDAR fusion projection

// Output formats

KITTInuScenesLASPCD
/ 1.6

NLP / Text Labeling

NER, intent classification, span labeling, RLHF preference data — annotation that goes beyond simple category labels.

The CEO of Acme CorpORG announced a new partnership with OpenAIORG in San FranciscoLOC on April 17, 2026DATE.
// NER · 4 entity types · CoNLL output

// Details

  • Named entity recognition (NER)
  • Intent & sentiment classification
  • Preference pairs for RLHF / DPO
  • Span annotation for SQuAD-style reading comprehension

// Output formats

CONLLJSONCustom schema
/ 1.7

Audio Transcription

ASR-grade transcription, speaker diarization, intent tagging for conversational AI.

// audio · diarized · 2 speakersSPK_01 · agentSPK_02 · customer[00:14] “Thanks for calling — how can I help today?”[00:18] “Yeah hi, I'd like to check my recent invoice…”

// Details

  • ASR training transcription at ≤ 5% WER
  • Speaker diarization across multi-party calls
  • Intent / entity tagging on utterances
  • Language support: English, Hindi, regional Indian languages

// Output formats

VTTSRTJSONCustom TSV
/ 1.8

Dataset Curation & Audit

Label audits, class-balance reports, dataset cleaning, and edge-case mining. The unglamorous work that changes model performance.

// audit_report · dataset_v3.tar · 14,820 items
class_balance ........ SKEWED car: 64% / person: 22% / sign: 14%
label_noise ......... 214 items inconsistent class assignment
tightness_iou ....... 0.89 mean ≥ benchmark
duplicates ......... 387 pairs perceptual hash match
edge_cases ......... 142 mined rain · night · occluded
resolution_drift ..... stable
action_items ......... 4 critical · 9 advisable

// Details

  • Full-dataset label audits with error categorization
  • Class-balance and distribution analysis
  • Edge-case mining with active learning support
  • Consistency checks across label rounds

// Output formats

Audit report PDFJSON diffCorrected dataset
// Work with us

Ready to ship? Let's scope it together.

Whether it's labeled data, a fine-tuned model, a RAG pipeline, or an agent running in production — bring us the brief. We'll scope it, price it, and tell you honestly if we're the right team. Inside 48 hours, no commitment.