evren-sdk 0.5.0

EVREN MLOps Platform — Python inference SDK for object detection, classification, and segmentation models.

evren-sdk

Türkçe · English

---

Türkçe

EVREN platformu üzerinde eğitilmiş bilgisayarlı görü modellerine Python'dan çıkarım (inference) yapmanızı sağlayan resmi SDK.

Desteklenen görevler: nesne tespiti, sınıflandırma, segmentasyon, döndürülmüş kutu (OBB), anahtar nokta (keypoint).

pip install evren-sdk

Hızlı Başlangıç

from evren_sdk import EvrenClient

client = EvrenClient(api_key="evren_xxxxx")

result = client.predict("kullanici/model-adi", "foto.jpg", confidence=0.3)

for det in result.predictions:
    print(f"{det.class_name}: {det.confidence:.0%}  bbox={det.bbox}")

print(f"Çıkarım süresi: {result.inference_ms:.0f} ms")

Kimlik Doğrulama

Platformda Ayarlar → API Anahtarları sayfasından anahtar oluşturun. Anahtar evren_ ön eki ile başlar ve X-API-Key başlığı üzerinden otomatik gönderilir.

# API anahtarı ile (önerilen)
client = EvrenClient(api_key="evren_xxxxx")

# JWT token ile de çalışır
client = EvrenClient(api_key="eyJhbGci...")

Tekil Çıkarım

result = client.predict(
    model="kullanici/model-adi",        # slug veya UUID
    image="resim.jpg",                  # dosya yolu, Path veya bytes
    confidence=0.25,
    iou=0.45,
    image_size=640,
    classes=["araba", "insan"],         # isteğe bağlı sınıf filtresi
)

model parametresi üç format kabul eder:

Format	Açıklama
"owner/slug"	Son versiyonu otomatik çözer
"owner/slug:v2.0"	Belirli versiyon etiketi
"019cec..." (UUID)	Doğrudan versiyon kimliği

Toplu Çıkarım (Batch)

Birden fazla görseli tek istekte GPU batch çıkarımı ile işleyin.

batch = client.predict_batch(
    model="kullanici/model-adi",
    images=["img1.jpg", "img2.jpg", "img3.jpg"],
    confidence=0.3,
)

for r in batch:
    print(f"{r.count} tespit, {r.inference_ms:.0f} ms")

print(f"Toplam: {batch.total_ms:.0f} ms, {batch.count} görsel")

Model Bilgileri

info = client.model_classes("kullanici/model-adi")
print(info.architecture)
for cls in info.classes:
    print(f"  {cls.name}: {cls.color}")

for m in client.list_models():
    print(f"{m.full_slug} — {m.architecture}")

Ön Yükleme (Warmup)

İlk çıkarım gecikmesini ortadan kaldırmak için modelleri önceden GPU'ya yükleyin.

client.warmup(["kullanici/model-adi", "diger/model"])

Asenkron Kullanım

import asyncio
from evren_sdk import AsyncEvrenClient

async def main():
    async with AsyncEvrenClient(api_key="evren_xxxxx") as client:
        result = await client.predict("kullanici/model-adi", "foto.jpg")
        batch = await client.predict_batch(
            "kullanici/model-adi", ["a.jpg", "b.jpg"],
        )

asyncio.run(main())

Hata Yönetimi

from evren_sdk import EvrenClient, NotFoundError, RateLimitError, InferenceError

client = EvrenClient(api_key="evren_xxxxx")

try:
    result = client.predict("kullanici/model", "test.jpg")
except NotFoundError:
    print("Model bulunamadı")
except RateLimitError as e:
    time.sleep(e.retry_after)
except InferenceError:
    print("GPU sunucusu geçici olarak kullanılamıyor")

Exception	HTTP	Açıklama
AuthenticationError	401, 403	Geçersiz veya süresi dolmuş anahtar
NotFoundError	404	Model veya versiyon bulunamadı
ValidationError	422	Hatalı parametre
RateLimitError	429	İstek limiti aşıldı
InferenceError	502, 503	GPU sunucusu hatası

Edge Modu (GPU'suz Cihazlar)

GPU olmayan cihazlarda (Raspberry Pi, laptop, endüstriyel PC) kamera veya video üzerinden gerçek zamanlı çıkarım yapın. Çıkarım EVREN GPU'larında çalışır, kullanıcı lokal model gibi deneyimler.

pip install evren-sdk[edge]

from evren_sdk import EvrenCamera

cam = EvrenCamera("evren_...", "owner/model")

# Tek satir: webcam ac, ESC ile kapat
cam.run(0)

# Iterator olarak — kendi dongunuzde kullanin
for frame, result in cam.stream(0):
    print(f"{result.count} tespit, {result.inference_ms:.0f}ms")
    # frame zaten annotated (bbox + label cizili)

# Video dosyasi isle + annotated cikti kaydet
cam.record("input.mp4", "output.mp4")

# Klasordeki gorselleri toplu isle
for path, result in cam.scan("images/", save_to="results/"):
    print(f"{path.name}: {result.count} nesne")

# RTSP IP kamera
for frame, result in cam.stream("rtsp://192.168.1.10/stream"):
    ...

Desteklenen kaynaklar: webcam (0, 1), video dosyası (*.mp4, *.avi), RTSP akışı, HTTP stream, görsel klasörü.

Parametre	Varsayılan	Açıklama
max_fps	15.0	Bant genişliğini korumak için FPS limiti
jpeg_quality	70	JPEG sıkıştırma kalitesi (20-95)
draw	True	Tahminleri frame üzerine çiz
confidence	0.25	Minimum güven eşiği

# draw_predictions() bagimsiz kullanilabilir
from evren_sdk import draw_predictions

frame = cv2.imread("foto.jpg")
result = client.predict("owner/model", "foto.jpg")
draw_predictions(frame, result.predictions)
cv2.imwrite("annotated.jpg", frame)

---

English

Official Python SDK for running inference on computer vision models trained on the EVREN platform.

Supported tasks: object detection, classification, segmentation, oriented bounding box (OBB), keypoint detection.

pip install evren-sdk

Quick Start

from evren_sdk import EvrenClient

client = EvrenClient(api_key="evren_xxxxx")

result = client.predict("owner/model-name", "photo.jpg", confidence=0.3)

for det in result.predictions:
    print(f"{det.class_name}: {det.confidence:.0%}  bbox={det.bbox}")

print(f"Inference time: {result.inference_ms:.0f} ms")

Authentication

Create an API key from Settings → API Keys on the platform. Keys start with the evren_ prefix and are sent automatically via the X-API-Key header.

# API key (recommended)
client = EvrenClient(api_key="evren_xxxxx")

# JWT tokens also work
client = EvrenClient(api_key="eyJhbGci...")

Single Prediction

result = client.predict(
    model="owner/model-name",           # slug or UUID
    image="image.jpg",                  # file path, Path, or bytes
    confidence=0.25,
    iou=0.45,
    image_size=640,
    classes=["car", "person"],          # optional class filter
)

The model parameter accepts three formats:

Format	Description
"owner/slug"	Resolves to latest version automatically
"owner/slug:v2.0"	Specific version tag
"019cec..." (UUID)	Direct version ID

Batch Prediction

Process multiple images in a single request with GPU batch inference.

batch = client.predict_batch(
    model="owner/model-name",
    images=["img1.jpg", "img2.jpg", "img3.jpg"],
    confidence=0.3,
)

for r in batch:
    print(f"{r.count} detections, {r.inference_ms:.0f} ms")

print(f"Total: {batch.total_ms:.0f} ms, {batch.count} images")

Model Information

info = client.model_classes("owner/model-name")
print(info.architecture)
for cls in info.classes:
    print(f"  {cls.name}: {cls.color}")

for m in client.list_models():
    print(f"{m.full_slug} — {m.architecture}")

Warmup

Pre-load models onto the GPU to eliminate cold-start latency.

client.warmup(["owner/model-name", "other/model"])

Async Usage

Same API surface with async/await:

import asyncio
from evren_sdk import AsyncEvrenClient

async def main():
    async with AsyncEvrenClient(api_key="evren_xxxxx") as client:
        result = await client.predict("owner/model-name", "photo.jpg")
        batch = await client.predict_batch(
            "owner/model-name", ["a.jpg", "b.jpg"],
        )

asyncio.run(main())

Error Handling

from evren_sdk import EvrenClient, NotFoundError, RateLimitError, InferenceError

client = EvrenClient(api_key="evren_xxxxx")

try:
    result = client.predict("owner/model", "test.jpg")
except NotFoundError:
    print("Model not found")
except RateLimitError as e:
    time.sleep(e.retry_after)
except InferenceError:
    print("GPU server temporarily unavailable")

Exception	HTTP	Description
AuthenticationError	401, 403	Invalid or expired key
NotFoundError	404	Model or version not found
ValidationError	422	Invalid parameter
RateLimitError	429	Rate limit exceeded
InferenceError	502, 503	GPU server error

Edge Mode (GPU-free Devices)

Run real-time inference on devices without a GPU (Raspberry Pi, laptops, industrial PCs). Inference runs on EVREN's cloud GPUs — the user experience feels completely local.

pip install evren-sdk[edge]

from evren_sdk import EvrenCamera

cam = EvrenCamera("evren_...", "owner/model")

# One-liner: opens webcam, press ESC to quit
cam.run(0)

# Iterator — use in your own loop
for frame, result in cam.stream(0):
    print(f"{result.count} detections, {result.inference_ms:.0f}ms")
    # frame is already annotated (bboxes + labels drawn)

# Process video file + save annotated output
cam.record("input.mp4", "output.mp4")

# Batch process a folder of images
for path, result in cam.scan("images/", save_to="results/"):
    print(f"{path.name}: {result.count} objects")

# RTSP IP camera
for frame, result in cam.stream("rtsp://192.168.1.10/stream"):
    ...

Supported sources: webcam (0, 1), video files (*.mp4, *.avi), RTSP streams, HTTP streams, image folders.

Parameter	Default	Description
max_fps	15.0	FPS cap to conserve bandwidth
jpeg_quality	70	JPEG compression quality (20-95)
draw	True	Render predictions on frame
confidence	0.25	Minimum confidence threshold

# draw_predictions() works standalone
from evren_sdk import draw_predictions

frame = cv2.imread("photo.jpg")
result = client.predict("owner/model", "photo.jpg")
draw_predictions(frame, result.predictions)
cv2.imwrite("annotated.jpg", frame)

Data Models

Class	Key Fields
PredictResult	predictions, inference_ms, image_width, image_height, count
Prediction	class_name, confidence, bbox, color, mask, keypoints, obb
BatchResult	results, total_ms, count — iterable
ModelClasses	classes, architecture, model_name, imgsz — supports in operator
ModelInfo	id, name, slug, architecture, full_slug
ModelVersion	id, version_tag, framework, metrics
EvrenCamera	stream(), run(), scan(), record(), stats

---

Requirements

• Python >= 3.10
• httpx >= 0.27
• opencv-python >= 4.8 (only for evren-sdk[edge])

License

Apache License 2.0