# Detect Objects You can use the Nvision’s **Object Detection** service to detect labels in an image. this service localizes and identifies multiple objects in the image such as people, animals, vehicles, and furniture. see [machine learning services](https://docs.nvision.nipa.cloud/machine-learning-services). ![](/files/-M4CTZ8dvHldDGpI436R) A response is returned in JSON format similar to the following example: ```javascript { "detected_objects": [ { "confidence": 0.8327999711036682, "parent": "accessory", "bounding_box": { "left": 246, "right": 303, "top": 520, "bottom": 605 }, "name": "backpack" }, ... { "confidence": 0.6195999979972839, "parent": "Object", "bounding_box": { "left": 595, "right": 641, "top": 64, "bottom": 230 }, "name": "traffic light" } ] } ``` {% hint style="info" %} If you **have not created a Nvision service** account credentials, do so now in this [set up the Nvision service](https://docs.nvision.nipa.cloud/quickstarts/set-up-the-nvision-api) quickstart for instructions. {% endhint %} Once your service has been created, go to the service overview page under API Key to get your service key. ## Detect objects in an image ### Image Content The Nvision API can perform object detection on a local image file by sending an image as a **base64 encoded** **string** in your request body. The [base64](https://en.wikipedia.org/wiki/Base64) encoded string is a binary-to-text encoding that represents binary data in an ASCII **string** format as the following example: `/9j/4AAQSkZJRgABAQEBLAEsAAD...` ### JSON Request Body The API is accessible via the HTTP method and URL, see : **`POST`** `https://nvision.nipa.cloud/api/v1/object-detection` ```javascript { "raw_data": <>, "configurations": [ { "parameter": "OutputCroppedImage", "value": "false" }, { "parameter": "ConfidenceThreshold", "value": "0.1" } ] } ``` The configuration is different on individual service types. It is structured as a key-value mapping. A config name is defined in **`parameter`** field and the corresponding value is defined in **`value`** field in **string format**. For object detection service, there are two available configurations as follows: * **`ConfidenceThreshold`**: to define the minimum confidence score of the prediction results. * Value options: `[0, 1]` * Default: `"0.1"` * **`OutputCroppedImage`**: to return cropped images from bounding box detections. * Value options: `"true"` or `"false"` * Default: `"false"` * **`OutputVisualizedImage`**: to return drawn bounding box detections on raw image. * Value options: `"true"` or `"false"` * Default: `"false"` ## Make a RESTful Call You can call this API through REST calls or native SDKs. ### **Using the cURL command line** ```bash export API_KEY="<>" # save the json request body as a file named request.json curl -X POST \ https://nvision.nipa.cloud/api/v1/object-detection -H 'Authorization: ApiKey '$API_KEY \ -H 'Content-Type: application/json' \ -d @request.json | json_pp # or read a local image from filepath echo -n '{"raw_data": "'"$(base64 image.jpg)"'"}' | \ curl -X POST \ https://nvision.nipa.cloud/api/v1/object-detection \ -H 'Authorization: ApiKey '$API_KEY \ -H "Content-Type: application/json" \ -d @- | json_pp ``` ### **Using the client libraries** Nvision SDKs provide interface for calling Nvision services in your own language. #### **Installation** * For python, using pypi package here: * For nodejs, using npm package here: {% tabs %} {% tab title="Python" %} ```bash pip install nvision ``` {% endtab %} {% tab title="JavaScript" %} ```bash yarn init yarn add @nipacloud/nvision ``` {% endtab %} {% endtabs %} #### Example Usage See the [SDK Reference](https://docs.nvision.nipa.cloud/api-references/python-sdk), in this guide, covers calling Nvision API for **Python** and **JavaScript**. {% tabs %} {% tab title="Python" %} ```python import os import json import base64 from nvision import ObjectDetection model = ObjectDetection(api_key='YOUR_API_KEY') # base64 encoed string with open('image.jpg', 'rb') as file: image = file.read() image = base64.b64encode(image).decode('utf-8') # make a RESTful call to the Nvision API response = model.predict(image) # get the predictions (in JSON format) from the response print(json.dumps(response.json(), indent=4, sort_keys=True)) ``` {% endtab %} {% tab title="JavaScript" %} ```typescript const nvision = require("@nipacloud/nvision"); const objectDetectionService = nvision.objectDetection({ apiKey: "" }); objectDetectionService.predict({ rawData: "BASE64_ENCODED_IMAGE" }).then((result) => { // Outout the result object to console console.log(result); }); ``` {% endtab %} {% endtabs %} **Set up the SDK with Webpack (optional)** If you use the SDK in the webpack-based project, you can provide the module resolution alias in your webpack configuration. {% tabs %} {% tab title="JavaScript" %} ```javascript { ... resolve: { alias: { "@nipacloud/nvision": "@nipacloud/nvision/dist/browser/nvision.js" } } ... } ``` {% endtab %} {% endtabs %} ## Detect Object in a video An example of how to integrate `@nipacloud/nvision` SDK to the frontend app.\ or see [making a websocket stream quickstart.](/quickstarts/make-a-websocket-stream.md) {% content-ref url="/pages/-M4CTYjPOObkwvzTcB4x" %} [Make a WebSocket Stream](/quickstarts/make-a-websocket-stream.md) {% endcontent-ref %} --- # Agent Instructions: Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://docs.nvision.nipa.cloud/how-to-guides/detect-objects.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.