armanid 1.0.0.17

Armanid — Object Detection Library by Epic Rabbit

Armanid

Awesome Real-time Machine Analysis for Next-gen Intelligent Detection

     

        

---

Overview

Armanid is a production-grade object detection engine designed for enterprises that demand accuracy, speed, and reliability — from edge devices to cloud-scale deployments.

Built by Epic Rabbit, Armanid delivers 99.7% mAP on the COCO dataset, real-time inference at 120 FPS, and a developer-first Python API that gets you from install to first inference in under five minutes.

---

Installation

# Standard install
pip install armanid

# With GPU acceleration (NVIDIA CUDA)
pip install armanid[cuda]

# Enterprise production suite
pip install armanid[enterprise]

# Full development environment
pip install armanid[dev]

---

Quick Start

from armanid import Armanid

# Initialize detector
model = Armanid("armanid-medium")

# Run on an image
results = model.detect("image.jpg", conf=0.85)

# Inspect results
for box in results[0].boxes:
    print(box.label, box.conf, box.xyxy)

# Save annotated output
results[0].save("output.jpg")

---

Model Family

Five precision tiers — from ultra-fast edge deployment to maximum-accuracy research workloads.

Model	Size	Speed	mAP	Use Case
armanid-nano	~6 MB	120 FPS	85%	Mobile / Edge
armanid-small	~22 MB	80 FPS	92%	Real-time
armanid-medium	~50 MB	45 FPS	95%	General
armanid-large	~84 MB	25 FPS	97%	High-quality
armanid-xlarge	~131 MB	15 FPS	98%	Research

Specialised variants also available: armanid-nano-seg (segmentation) and armanid-nano-pose (pose estimation).

---

API Reference

Detection

from armanid import Armanid

model = Armanid("armanid-large")

# Image detection
results = model.detect("image.jpg", conf=0.25)

# Filter by class — 0: person, 2: car, 5: bus
results = model.detect(
    source="rtsp://your-camera",
    conf=0.85,
    iou=0.65,
    classes=[0, 2, 5]
)

# Result accessors
result = results[0]
result.boxes          # List of BoundingBox objects
result.labels         # ["person", "car", ...]
result.confidences    # [0.92, 0.87, ...]
result.summary()      # {"person": 2, "car": 1}
result.show()         # Display annotated image
result.save("out.jpg")
result.plot()         # Returns annotated numpy array

Bounding Box Fields

box = result.boxes[0]

box.label      # "person"
box.conf       # 0.92
box.xyxy       # (x1, y1, x2, y2)
box.xywh       # (cx, cy, w, h)
box.width
box.height
box.area
box.track_id   # Available when tracking

Object Tracking

from armanid import Armanid

model = Armanid("armanid-large")

# Stream video with persistent tracking
for result in model.track("video.mp4", stream=True):
    for box in result.boxes:
        print(box.track_id, box.label, box.conf)

# Live webcam with FPS overlay
from armanid.utils import run_webcam
run_webcam(model, conf=0.8, show_fps=True)

Custom Training

from armanid import Armanid

# Load base model
model = Armanid("armanid-large.yaml")

# Fine-tune on custom dataset
model.train(
    data="custom-dataset.yaml",
    epochs=100,
    batch_size=16,
    optimizer="adamw"
)

# Export for production
model.export(format="onnx")

Enterprise Deployment

# Docker
docker run -p 8080:8080 armanid/enterprise:latest

# Kubernetes
kubectl apply -f armanid-deployment.yaml

# REST API
curl -X POST "http://localhost:8080/detect" \
     -H "Content-Type: application/json" \
     -d '{"image": "base64_encoded_image"}'

---

Performance

Speed (FPS)

armanid-nano	120 FPS
armanid-small	80 FPS
armanid-medium	45 FPS
armanid-large	25 FPS
armanid-xlarge	15 FPS

Accuracy (mAP on COCO)

armanid-nano	85%
armanid-small	92%
armanid-medium	95%
armanid-large	97%
armanid-xlarge	98%

---

System Requirements

Minimum

Requirement	Value
Python	3.8+
RAM	4 GB
Storage	500 MB
OS	Windows 10+, Ubuntu 18.04+, macOS 10.15+

Recommended (Enterprise)

Requirement	Value
Python	3.10+
RAM	16 GB+
GPU	NVIDIA RTX 30-series or newer
Storage	50 GB SSD
OS	Ubuntu 20.04 LTS

---

Industries

Armanid is deployed in production across six sectors where precision and uptime are non-negotiable.

• Manufacturing — Quality control and defect detection. Reduces inspection time by up to 85%.
• Automotive — ADAS and autonomous driving perception pipelines.
• Healthcare — Medical imaging analysis compatible with DICOM workflows.
• Security — Multi-camera surveillance and real-time threat detection.
• Retail — Inventory management, shelf analytics, and customer flow tracking.
• Construction — PPE compliance, safety monitoring, and project progress tracking.

---

Support

Channel	Link
Documentation	docs.armanid.ai
Community	discord.armanid.ai
Enterprise Email	[e-mail]
Tutorials	YouTube Channel

Enterprise plans include a dedicated success manager, 24/7 technical support, custom model training, performance optimization, and SLA guarantees.

---

About

Built by Epic Rabbit — creators of next-generation AI solutions.

Arman Guriyan — Founder & CEO  |  LinkedIn  |  [e-mail]

Licensed under the GNU Affero General Public License v3.0 (AGPL-3.0).
Copyright © 2025 Arman Guriyan — Epic Rabbit.