Uni Har

Multi-Scale Multimodal Pose HAR

A custom PyTorch model for Human Activity Recognition (HAR) that integrates:

• Short-term pose transformers (factorized temporal + spatial attention)
• Long-term temporal aggregation
• Optional multimodal fusion with RGB images
• Multi-stage self-supervised + supervised training pipeline

---

🧠 Model Overview

Architecture Summary

Pose stream

• Input: (B, L, T, J, C)
• Short-term encoder: PoseFormerFactorized
  • Temporal attention (per joint)
  • Spatial attention (per frame)
• Long-term encoder: Transformer over segment-level features

Image stream (optional)

• Backbone: ResNet18 / ResNet50
• Temporal pooling per segment

Fusion

• concat (default): feature concatenation + MLP
• xattn: shallow cross-attention (pose tokens ↔ image token)

Output

• Activity classification logits
• Optional intermediate embeddings / tokens

---

🏗️ Model Components

Module	Description
PoseFormerFactorized	Short-term pose transformer
LongTermTemporalBlock	Long-range temporal modeling
ImageEncoder	CNN-based RGB feature extractor
MMFusionConcatLN	Concatenation-based multimodal fusion
MMFusionCrossAttnShallow	Cross-attention multimodal fusion
SSLHeads	Contrastive, reconstruction, temporal order heads

---

📥 Input Format

Pose Input

(B, L, T, J, C)

• B: batch size
• L: number of temporal segments
• T: frames per segment
• J: number of joints (e.g. 17)
• C: joint channels (2D or 3D)

Image Input (optional)

(B, L, T, 3, H, W)

• 3: RGB channels
• H: image height
• W: image width

---

🚀 Usage

1️⃣ Load Model Code

from model_har_final import (
    PoseFormerFactorized,
    MultiScaleTemporalModel
)

---

2️⃣ Build Model

pose_backbone = PoseFormerFactorized(
    joints=17,
    in_ch=3,
    dim=128,
    layers=4,
    num_classes=6,
    return_tokens=True
)

model = MultiScaleTemporalModel(
    short_seq_model=pose_backbone,
    num_classes=6,
    enable_long_term=True,
    multimodal=True,
    fusion_mode="concat"  # or "xattn"
)

---

3️⃣ Load Weights

import torch

ckpt = torch.load("best_stage3_dual_sched.pth", map_location="cpu")
model.load_state_dict(ckpt)
model.eval()

i️ This model is saved using state_dict, not pickle-serialized objects, for maximum compatibility.

---

4️⃣ Inference

with torch.no_grad():
    logits = model(pose_seq, img_seq)
    preds = logits.argmax(dim=-1)

---

🧪 Training Strategy

The model is trained in three stages:

Stage 1 — Pose SSL + Weak Supervision

• Masked joint modeling (MJM)
• Contrastive learning (InfoNCE)
• Temporal order prediction
• Optional labeled supervision

Stage 2 — Pose-only SSL Refinement

• Contrastive + reconstruction losses
• Temporal attention disabled for stability

Stage 3 — Multimodal Supervised Fine-tuning

• Label smoothing CE
• Semantic prototype distillation
• Metric learning (triplet)
• Optional knowledge distillation

---

📊 Outputs

Output	Shape
Logits	(B, num_classes)
Pose embedding	(B, D)
Pose tokens (optional)	(B, T, J, D)

---

📎 Limitations

• Not compatible with AutoModel.from_pretrained
• Requires custom code to instantiate architecture
• Input pose format must match training configuration

---

📜 License

Apache License 2.0

---

📌 Citation

If you use this work, please cite:

@misc{kim2025multiscalehar,
  title        = {Multi-Scale Multimodal Pose Transformer for Human Activity Recognition},
  author       = {Minjae Kim},
  year         = {2025},
  howpublished = {\url{https://huggingface.co/m97j/har-safety-model}}
}