Skip to content

HTTP Inference

The Roboflow Inference Server provides a standard API through which to run inference on computer vision models.

In this guide, we show how to run inference on object detection, classification, and segmentation models using the Inference Server.

Currently, the server is compatible with models trained on Roboflow, but stay tuned as we actively develop support for bringing your own models.

To run inference with the server, we will:

  1. Install the server
  2. Download a model for use on the server
  3. Run inference

Step #1: Install the Inference Server

You can skip this step if you already have Inference installed and running.

The Inference Server runs in Docker. Before we begin, make sure you have installed Docker on your system. To learn how to install Docker, refer to the official Docker installation guide.

Once you have Docker installed, you are ready to download Roboflow Inference. The command you need to run depends on what device you are using.

Info

Prior to installation, you may want to configure a python virtual environment to isolate dependencies of inference.

To install Inference via pip:

pip install inference

If you have an NVIDIA GPU, you can accelerate your inference with:

pip install inference-gpu

Start the server using inference server start. After you have installed the Inference Server, the Docker container will start running the server at localhost:9001.

Step #2: Run Inference

You can send a URL with an image, a NumPy array, or a base64-encoded image to an Inference server. The server will return a JSON response with the predictions.

There are two generations of routes in a Roboflow inference server To see what routes are available for a running inference server instance, visit the /docs route in a browser. Roboflow hosted inference endpoints (detect.roboflow.com) only support V1 routes.

Run Inference on a v2 Route

Run

Create a new Python file and add the following code:

import requests

project_id = "soccer-players-5fuqs"
model_version = 1
image_url = "https://storage.googleapis.com/com-roboflow-marketing/inference/soccer.jpg"
confidence = 0.75
iou_thresh = 0.5
api_key = "YOUR API KEY"
task = "object_detection"

infer_payload = {
    "model_id": f"{project_id}/{model_version}",
    "image": {
        "type": "url",
        "value": image_url,
    },
    "confidence": confidence,
    "iou_threshold": iou_thresh,
    "api_key": api_key,
}
res = requests.post(
    f"http://localhost:9001/infer/{task}",
    json=infer_payload,
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. image_url: The URL of the image you want to run inference on.
  3. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  4. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  5. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.

Then, run the Python script:

python app.py

Create a new Python file and add the following code:

import requests
import base64
from PIL import Image
from io import BytesIO

project_id = "soccer-players-5fuqs"
model_version = 1
task = "object_detection"
confidence = 0.5
iou_thresh = 0.5
api_key = "YOUR ROBOFLOW API KEY"
file_name = "path/to/local/image.jpg"

image = Image.open(file_name)

buffered = BytesIO()

image.save(buffered, quality=100, format="JPEG")

img_str = base64.b64encode(buffered.getvalue())
img_str = img_str.decode("ascii")

infer_payload = {
    "model_id": f"{project_id}/{model_version}",
    "image": {
        "type": "base64",
        "value": img_str,
    },
    "confidence": confidence,
    "iou_threshold": iou_thresh,
    "api_key": api_key,
}

res = requests.post(
    f"http://localhost:9001/infer/{task}",
    json=infer_payload,
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  3. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  4. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.
  5. filename: The path to the image you want to run inference on.

Then, run the Python script:

python app.py

Object detection models support batching. Utilize batch inference by passing a list of image objects in a request payload:

import requests
import base64
from PIL import Image
from io import BytesIO

project_id = "soccer-players-5fuqs"
model_version = 1
task = "object_detection"
confidence = 0.5
iou_thresh = 0.5
api_key = "YOUR ROBOFLOW API KEY"
file_name = "path/to/local/image.jpg"

image = Image.open(file_name)

buffered = BytesIO()

image.save(buffered, quality=100, format="JPEG")

img_str = base64.b64encode(buffered.getvalue())
img_str = img_str.decode("ascii")

infer_payload = {
    "model_id": f"{project_id}/{model_version}",
    "image": [
        {
            "type": "base64",
            "value": img_str,
        },
        {
            "type": "base64",
            "value": img_str,
        },
        {
            "type": "base64",
            "value": img_str,
        }
    ],
    "confidence": confidence,
    "iou_threshold": iou_thresh,
    "api_key": api_key,
}

res = requests.post(
    f"http://localhost:9001/infer/{task}",
    json=infer_payload,
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  3. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  4. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.
  5. filename: The path to the image you want to run inference on.

Then, run the Python script:

python app.py

Create a new Python file and add the following code:

import requests
import base64
from PIL import Image
from io import BytesIO

project_id = "soccer-players-5fuqs"
model_version = 1
task = "object_detection"
confidence = 0.5
iou_thresh = 0.5
api_key = "YOUR ROBOFLOW API KEY"
file_name = "path/to/local/image.jpg"

image = cv2.imread(file_name)
numpy_data = pickle.dumps(image)
img_str = base64.b64encode(numpy_data)
img_str = img_str.decode("ascii")

infer_payload = {
    "model_id": f"{project_id}/{model_version}",
    "image": {
        "type": "numpy",
        "value": img_str,
    },
    "confidence": confidence,
    "iou_threshold": iou_thresh,
    "api_key": api_key,
}

res = requests.post(
    f"http://localhost:9001/infer/{task}",
    json=infer_payload,
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  3. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  4. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.
  5. filename: The path to the image you want to run inference on.

Then, run the Python script:

python app.py

Run Inference on a v1 Route

Run

The Roboflow hosted API uses the V1 route and requests take a slightly different form:

```python
import requests
import base64
from PIL import Image
from io import BytesIO

project_id = "soccer-players-5fuqs"
model_version = 1
confidence = 0.5
iou_thresh = 0.5
api_key = "YOUR ROBOFLOW API KEY"
image_url = "https://storage.googleapis.com/com-roboflow-marketing/inference/soccer.jpg


res = requests.post(
    f"https://detect.roboflow.com/{project_id}/{model_version}?api_key={api_key}&confidence={confidence}&overlap={iou_thresh}&image={image_url}",
)

predictions = res.json()
print(predictions)
```

Above, specify:

1. `project_id`, `model_version`: Your project ID and model version number. <a href="https://docs.roboflow.com/api-reference/workspace-and-project-ids" target="_blank">Learn how to retrieve your project ID and model version number</a>.
2. `confidence`: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
3. `api_key`: Your Roboflow API key. <a href="https://docs.roboflow.com/api-reference/authentication#retrieve-an-api-key" target="_blank">Learn how to retrieve your Roboflow API key</a>.
4. `task`: The type of task you want to run. Choose from `object_detection`, `classification`, or `segmentation`.
5. `filename`: The path to the image you want to run inference on.

Then, run the Python script:

```
python app.py
```

The Roboflow hosted API uses the V1 route and requests take a slightly different form:

import requests
import base64
from PIL import Image
from io import BytesIO

project_id = "soccer-players-5fuqs"
model_version = 1
confidence = 0.5
iou_thresh = 0.5
api_key = "YOUR ROBOFLOW API KEY"
file_name = "path/to/local/image.jpg"

image = Image.open(file_name)

buffered = BytesIO()

image.save(buffered, quality=100, format="JPEG")

img_str = base64.b64encode(buffered.getvalue())
img_str = img_str.decode("ascii")

res = requests.post(
    f"https://detect.roboflow.com/{project_id}/{model_version}?api_key={api_key}&confidence={confidence}&overlap={iou_thresh}",
    data=img_str,
    headers={"Content-Type": "application/json"},
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  3. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  4. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.
  5. filename: The path to the image you want to run inference on.

Then, run the Python script:

python app.py

Numpy arrays can be pickled and passed to the inference server for quicker processing. Note, Roboflow hosted APIs to not accept numpy inputs for security reasons:

import requests
import cv2
import pickle

project_id = "soccer-players-5fuqs"
model_version = 1
task = "object_detection"
api_key = "YOUR API KEY"
file_name = "path/to/local/image.jpg"

image = cv2.imread(file_name)
numpy_data = pickle.dumps(image)

res = requests.post(
    f"http://localhost:9001/{project_id}/{model_version}?api_key={api_key}&image_type=numpy",
    data=numpy_data,
    headers={"Content-Type": "application/json"},
)

predictions = res.json()
print(predictions)

Above, specify:

  1. project_id, model_version: Your project ID and model version number. Learn how to retrieve your project ID and model version number.
  2. confidence: The confidence threshold for predictions. Predictions with a confidence score below this threshold will be filtered out.
  3. api_key: Your Roboflow API key. Learn how to retrieve your Roboflow API key.
  4. task: The type of task you want to run. Choose from object_detection, classification, or segmentation.
  5. filename: The path to the image you want to run inference on.

Then, run the Python script:

python app.py

Batch inference is not currently supported by V1 routes.

The code snippets above will run inference on a computer vision model. On the first request, the model weights will be downloaded and set up with your local inference server. This request may take some time depending on your network connection and the size of the model. Once your model has downloaded, subsequent requests will be much faster.

The Inference Server comes with a /docs route at localhost:9001/docs or localhost:9001/redoc that provides OpenAPI-powered documentation. You can use this to reference the routes available, and the configuration options for each route.

Batching Requests

Object detection models trained with Roboflow support batching, which allow you to upload multiple images of any type at once:

infer_payload = {
    "model_id": f"{project_id}/{model_version}",
    "image": [
        {
            "type": "url",
            "value": image_url_1,
        },
        {
            "type": "url",
            "value": image_url_2,
        },
        {
            "type": "url",
            "value": image_url_3,
        },
    ],
    "confidence": confidence,
    "iou_threshold": iou_thresh,
    "api_key": api_key,
}