Image Classification

Mobile Net¶

MobileNet V2 is a lightweight convolutional neural network. On the basis of MobileNet, it has made two major improvements: Inverted Residuals and Linear bottlenecks. The PaddleHub Module is trained on Baidu's self-built animal dataset and can be used for image classification and feature extraction. Currently, it supports the classification and recognition of 7978 animals. Details of the model can be found in the paper.

Reference：https://github.com/PaddlePaddle/PaddleHub/blob/release/v2.2/modules/image/face_detection/pyramidbox_lite_server

Sample result example¶

Enter the file path and the model will give its predictions：

Prerequisite¶

1. Dependencies¶

Please visit dependencies

2. mobilenet_v2_animals Prerequisites¶

3. Install the Model¶

hub install pyramidbox_lite_server

Serve the Model¶

Install Pinferencia¶

First, let's install Pinferencia.

pip install "pinferencia[streamlit]"

Create app.py¶

Let's save our predict function into a file app.py and add some lines to register it.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

import base64

import cv2
import numpy as np
import paddlehub as hub

from pinferencia import Server, task

classifier = hub.Module(name="mobilenet_v2_animals")


def base64_str_to_cv2(base64_str: str) -> np.ndarray:
    return cv2.imdecode(
        np.fromstring(base64.b64decode(base64_str), np.uint8), cv2.IMREAD_COLOR
    )


def predict(data: list) -> list:
    images = [base64_str_to_cv2(base64_img_str) for base64_img_str in data]
    return classifier.classification(images=images)


service = Server()
service.register(
    model_name="classifier",
    model=predict,
    metadata={"task": task.IMAGE_TO_TEXT},
)

Run the service, and wait for it to load the model and start the server:

$ uvicorn app:service --reload
INFO:     Uvicorn running on http://127.0.0.1:8000 (Press CTRL+C to quit)
INFO:     Started reloader process [xxxxx] using statreload
INFO:     Started server process [xxxxx]
INFO:     Waiting for application startup.
INFO:     Application startup complete.

$ pinfer app:service --reload

Pinferencia: Frontend component streamlit is starting...
Pinferencia: Backend component uvicorn is starting...

Test the service¶

curl --location --request POST \
    'http://127.0.0.1:8000/v1/models/classifier/predict' \
    --header 'Content-Type: application/json' \
    --data-raw '{"data": ["/9j/4AAQS........"]}'

{
    "model_name": "classifier",
    "data": [
        {
            "松鼠": 0.9506056308746338
        }
    ]
}

1
2
3
4
5
6
7
8
9

import requests


response = requests.post(
    url="http://localhost:8000/v1/models/classifier/predict",
    headers = {"Content-type": "application/json"},
    json={"data": ["/9j/4AAQS........"]},
)
print(response.json())

$ python test.py
{
    "model_name": "classifier",
    "data": [
        {
            "松鼠": 0.9506056308746338
        }
    ]
}

Even cooler, go to http://127.0.0.1:8000, and you will have a full documentation of your APIs.

You can also send predict requests just there!