Object Detection
YOLOv10
OpenVINO
PyTorch
GGUF
Safetensors
English
phi3
text-generation
disaster-response
Instructions to use sohail-kustagi/MAAS-Disaster-Response with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- YOLOv10
How to use sohail-kustagi/MAAS-Disaster-Response with YOLOv10:
from ultralytics import YOLOvv10 model = YOLOvv10.from_pretrained("sohail-kustagi/MAAS-Disaster-Response") source = 'http://images.cocodataset.org/val2017/000000039769.jpg' model.predict(source=source, save=True) - Notebooks
- Google Colab
- Kaggle
File size: 603 Bytes
e3b6c9f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 | # MAAS-LLM: End-to-End Video Benchmark
## Test Configuration
- **Video:** Pella Home Fire Drone Footage (720p)
- **Environment:** Local PC (x86 CPU Baseline)
- **Vision Model:** YOLOv10 (PyTorch CPU)
- **Reasoning Model:** Phi-3 4k Instruct (llama.cpp)
## Stage 1: Vanilla Baseline (No LoRA)
- **Vision Speed (YOLO):** 8.77 FPS
- **LLM Reasoning Latency:** 26.19s per command
## Stage 2: Fine-Tuned LoRA (Disaster Analyst)
- **Vision Speed (YOLO):** 10.33 FPS
- **LLM Reasoning Latency:** 29.86s per command
---
*Generated automatically by MAAS-LLM Evaluator for the Arm AI Optimization Challenge.*
|