---
language:
  - sw
tags:
  - sign-language
  - pose-landmarks
  - mediapipe
  - transformer
  - african-sign-language
  - KSL
  - isolated-sign-recognition
license: apache-2.0
datasets:
  - luciayen/KSL-Hand-Landmarks
metrics:
  - accuracy
model-index:
  - name: AfriSignEncoder Exp1 — KSL Baseline
    results:
      - task:
          type: video-classification
          name: Isolated Sign Word Recognition
        dataset:
          name: KSL-Hand-Landmarks
          type: luciayen/KSL-Hand-Landmarks
        metrics:
          - type: accuracy
            value: 1.0
            name: Validation Accuracy
---

# AfriSignEncoder Exp1 — KSL Baseline (LandmarkTransformer)

Part of the **AfriSignEncoder** research project: a multilingual African sign language recognition benchmark.
This checkpoint is the **Experiment 1 single-language baseline** for Kenyan Sign Language (KSL).

> **Scope note:** This baseline covers only 4 KSL glosses (father, hello, is, my) from a small
> demo-scale dataset. The 100% accuracy reflects the simplicity of the 4-class task and should
> not be generalised to full KSL vocabulary recognition.

## Model Description

Same LandmarkTransformer architecture as the CASL baseline, with a 4-class head.

| Component | Value |
|---|---|
| Architecture | LandmarkTransformer (custom) |
| Input | (B, 64, 225) float32 — 75 keypoints × 3 coords per frame |
| Embedding dim | 256 |
| Attention heads | 8 |
| Encoder layers | 4 |
| Feed-forward dim | 1,024 |
| Positional encoding | Learned |
| Classification head | Linear 256 → 4 |
| Parameters | ~3.24 M |

## Dataset

**KSL Hand Landmarks** — 4 Kenyan Sign Language words.

| Split | Samples | Classes |
|---|---|---|
| Train | 694 | 4 |
| Test | 124 | 4 |

Source: Kaggle `joanwachuka/ksl-hand-landmarks` → parquet at `luciayen/KSL-Hand-Landmarks`.

**Landmark caveat:** The original `.npy` files contain MediaPipe **Hands**-only keypoints
(42 joints × 3 = 126D). These are placed in dimensions [0:126]; dimensions [126:225]
(the 99 pose body dims) are zero-padded. The model therefore learns from hand shape
and motion only; the padded dimensions are always zero and contribute no signal.

**Data leakage fix:** The Kaggle archive contains two sub-directories — `data_split/`
(official train/val/test) and `dataset3/` (pre-split source). All 124 test samples appear
verbatim in `dataset3/`. The upload script excludes `dataset3/` entirely, giving a clean
train=694 / test=124 split with zero overlap.

## Training

Identical protocol to the CASL baseline.

| Setting | Value |
|---|---|
| Optimiser | AdamW (lr=3e-4, wd=1e-4) |
| LR schedule | OneCycleLR cosine |
| Max epochs | 60 |
| Batch size | 64 |
| Loss | CrossEntropy + label_smoothing=0.1 |
| Early stopping | patience=12 on val acc |
| Normalisation | Per-feature z-score (stats stored in checkpoint) |

## Results

| Metric | Value |
|---|---|
| Best validation accuracy | **100%** |
| Best checkpoint epoch | 3 |
| Final epoch (early stop) | 15 |
| Number of classes | 4 |

The 100% result is expected: 4 highly phonologically distinct signs, ~173 training samples
per class, and a well-regularised model. This result validates the pipeline; it does not
benchmark KSL at meaningful scale.

## Checkpoint Contents

```python
import torch
ck = torch.load("pytorch_model.bin", map_location="cpu")
# Keys: epoch, val_acc, model (state_dict), l2i (label→index dict),
#        mean (tensor 225,), std (tensor 225,)
# l2i = {"father": 0, "hello": 1, "is": 2, "my": 3}
```

## Limitations

- Only 4 classes — not a meaningful KSL benchmark.
- Pose dims are always zero (hand-only source data). Re-extraction with MediaPipe Holistic is planned.
- Small dataset; results will change with more data.

## Citation / Project

AfriSignEncoder research project, CMU, 2026. GitHub: `africansl_encoder`.