solidSF / Jesse / Weightless OCR Engine ZERO-WEIGHT MULTI-DIGIT OCR
AUTOROUTER ⚡ SILICON (OPENROAD) ⚡ RESEARCH WHITEPAPER ↗ JESSE CHAT
PIPELINE: 8-Way Connected Component Labeling + Stroke Merging + Dual-Valley Ligature Splitting + COM Normalization.
INTERACTIVE STROKE CANVAS
560×280 PX
VERTICAL PROJECTION PROFILE H(x) SPLIT: AUTO
STROKE GEOMETRY INVARIANTS
0 px
Detected Digits Count 0
Bounding Box Spans --
Aspect Ratio (W/H) 0.00
Center of Mass (14.0, 14.0)
Segmentation Latency 0.00 µs
DIGIT CROP & RECEPTIVE FIELD PREVIEWS
Draw strokes or select a preset like 69 or 420 to view segmented digit previews.
DETERMINISTIC INFERENCE RESULTS
$0.00 COST
-
DRAW ON CANVAS
--%
Δ = 0.00
BAYESIAN LOG-ODDS SURPLUS DISTRIBUTION
OFFICIAL BENCHMARK SCORECARD
VERIFIED RECEIPTS
1-Digit Test Accuracy 93.89%
Approach 1 (Segmentation) 88.20%
Approach 2 (Saccadic Viterbi) 86.60%
Approach 3 (Joint 100-Class) 79.50%
Cat/Dog Sketch (Pointy vs Floppy) 100.00%
Cat/Dog Photo (CIFAR-10 10k) 63.80% (Minchinton)
Cat/Dog Inference Latency 8.04 µs (Sketch) / 67 µs (Photo)
ASME Y14.5 GD&T (14 Classes) 99.93% (1399/1400)
GD&T Inference Latency <20 µs (Web) / 0 Weights
Multi-Digit Arbitrary Strings 100% (Clean Exemplars)
Floating-Point Weights 0 (Pure Integer Memory)
Streaming Training Time 0.195 s (CPU)
Per-Digit Inference Latency 5.10 µs
Inference Operational Cost $0.00 (Zero Token API)
ACTIVE RECEPTIVE TUPLES (MAP)
0 / 160 active
Draw strokes on the canvas or load a preset sample to inspect active RAM address lines.
WEIGHTLESS OCR SPECIFICATION

1. Spatial N-Tuple Addressing: Pixels are mapped into discrete binary address lines feeding direct RAM lookup tables, avoiding dense floating-point matrix multiplications.

2. Multi-Digit Stroke Merging: 8-way connected components merge proximate strokes along the horizontal axis, and split touching ligatures at vertical projection minima.

3. Center-of-Mass Invariant: Aspect-ratio preserving $20\times 20$ scaling and center-of-mass translation map any handwritten numeral onto canonical LeCun coordinate space.

4. Empirical Bayesian Believing: Unlike backpropagation, Jesse updates memory tables in a single streaming pass ($O(1)$ updates), achieving sub-20 microsecond execution on edge hardware.

DOCUMENT SPECIFICATION
CAD / DLA
PRECISION CAD PRESETS:
DOCUMENT LAYOUT ANALYSIS LAYERS
4 ACTIVE
TOKEN & DIMENSION SEARCH
ALL
ZERO-WEIGHT TELEMETRY
$0.00 COST
Extracted Tokens 25 tokens
Total Glyphs 122 glyphs
Total OCR Latency 4.12 ms
Per-Glyph Latency 5.10 µs
Perception Mode Deterministic RAM
Cloud API Calls 0 (100% Client RAM)
CAD DRAWING VIEWPORT — SSF-1000 PIVOT ASSEMBLY
1 / 1
DROP PDF OR CAD DRAWING IMAGE HERE
TOKEN INSPECTOR & EXTRACTION
STRUCTURED DATA
Ø.7500 ±.0006
DIMENSION / GD&T
Bounding Box [x,y,w,h] [440, 500, 78, 12]
Confidence Score 99.9%
Inference Latency 66.3 µs (13 glyphs)
ACTIVE 10-BIT RAM TUPLES:
TYPE CALLOUT NOMINAL TOL
# PART NO DESCRIPTION QTY