fal-ai/sam-3/image

SAM 3 is a unified foundation model for promptable segmentation in images and videos. It can detect, segment, and track objects using text or visual prompts such as points, boxes, and masks.

Inference

Commercial use

Schema

LLMs

Playground API

About

Segment Image

1. Calling the API#

Install the client#

The client provides a convenient way to interact with the model API.

npm install --save @fal-ai/client

Migrate to @fal-ai/client

The @fal-ai/serverless-client package has been deprecated in favor of @fal-ai/client. Please check the migration guide for more information.

Setup your API Key#

Set FAL_KEY as an environment variable in your runtime.

export FAL_KEY="YOUR_API_KEY"

Submit a request#

The client API handles the API submit protocol. It will handle the request status updates and return the result when the request is completed.

import { fal } from "@fal-ai/client";

const result = await fal.subscribe("fal-ai/sam-3/image", {
  input: {
    image_url: "https://raw.githubusercontent.com/facebookresearch/segment-anything-2/main/notebooks/images/truck.jpg"
  },
  logs: true,
  onQueueUpdate: (update) => {
    if (update.status === "IN_PROGRESS") {
      update.logs.map((log) => log.message).forEach(console.log);
    }
  },
});
console.log(result.data);
console.log(result.requestId);

Real-time via WebSockets#

This model has a real-time mode via websockets, this is supported via the fal.realtime client.

import { fal } from "@fal-ai/client";

const connection = fal.realtime.connect("fal-ai/sam-3/image", {
  onResult: (result) => {
    console.log(result);
  },
  onError: (error) => {
    console.error(error);
  },
  // Fetch short-lived JWT token from your backend
  tokenProvider: async (app) => {
    const response = await fetch("/api/fal/realtime-token", {
      method: "POST",
      headers: { "Content-Type": "application/json" },
      body: JSON.stringify({ app }),
    });
    return response.text();
  },
  tokenExpirationSeconds: 10,
});

connection.send({
  image_url: "https://raw.githubusercontent.com/facebookresearch/segment-anything-2/main/notebooks/images/truck.jpg"
});

2. Authentication#

The API uses an API Key for authentication. It is recommended you set the FAL_KEY environment variable in your runtime when possible.

API Key#

In case your app is running in an environment where you cannot set environment variables, you can set the API Key manually as a client configuration.

import { fal } from "@fal-ai/client";

fal.config({
  credentials: "YOUR_FAL_KEY"
});

Protect your API Key

When running code on the client-side (e.g. in a browser, mobile app or GUI applications), make sure to not expose your FAL_KEY. Instead, use a server-side proxy to make requests to the API. For more information, check out our server-side integration guide.

3. Queue#

Long-running requests

For long-running requests, such as training jobs or models with slower inference times, it is recommended to check the Queue status and rely on Webhooks instead of blocking while waiting for the result.

Submit a request#

The client API provides a convenient way to submit requests to the model.

import { fal } from "@fal-ai/client";

const { request_id } = await fal.queue.submit("fal-ai/sam-3/image", {
  input: {
    image_url: "https://raw.githubusercontent.com/facebookresearch/segment-anything-2/main/notebooks/images/truck.jpg"
  },
  webhookUrl: "https://optional.webhook.url/for/results",
});

Fetch request status#

You can fetch the status of a request to check if it is completed or still in progress.

import { fal } from "@fal-ai/client";

const status = await fal.queue.status("fal-ai/sam-3/image", {
  requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b",
  logs: true,
});

Get the result#

Once the request is completed, you can fetch the result. See the Output Schema for the expected result format.

import { fal } from "@fal-ai/client";

const result = await fal.queue.result("fal-ai/sam-3/image", {
  requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b"
});
console.log(result.data);
console.log(result.requestId);

4. Files#

Some attributes in the API accept file URLs as input. Whenever that's the case you can pass your own URL or a Base64 data URI.

Data URI (base64)#

You can pass a Base64 data URI as a file input. The API will handle the file decoding for you. Keep in mind that for large files, this alternative although convenient can impact the request performance.

Hosted files (URL)#

You can also pass your own URLs as long as they are publicly accessible. Be aware that some hosts might block cross-site requests, rate-limit, or consider the request as a bot.

Uploading files#

We provide a convenient file storage that allows you to upload files and use them in your requests. You can upload files using the client API and use the returned URL in your requests.

import { fal } from "@fal-ai/client";

const file = new File(["Hello, World!"], "hello.txt", { type: "text/plain" });
const url = await fal.storage.upload(file);

Auto uploads

The client will auto-upload the file for you if you pass a binary object (e.g. File, Data).

Read more about file handling in our file upload guide.

5. Schema#

Input#

image_url string* required

URL of the image to be segmented

prompt string

Text prompt for segmentation Default value: "wheel"

point_prompts list<PointPrompt>

List of point prompts

box_prompts list<BoxPrompt>

Box prompt coordinates (x_min, y_min, x_max, y_max). Multiple boxes supported - use object_id to group boxes for the same object or leave empty for separate objects.

apply_mask boolean

Apply the mask on the image. Default value: true

sync_mode boolean

If True, the media will be returned as a data URI.

output_format OutputFormatEnum

The format of the generated image. Default value: "png"

Possible enum values: jpeg, png, webp

return_multiple_masks boolean

If True, upload and return multiple generated masks as defined by max_masks.

max_masks integer

Maximum number of masks to return when return_multiple_masks is enabled. Default value: 3

include_scores boolean

Whether to include mask confidence scores.

include_boxes boolean

Whether to include bounding boxes for each mask (when available).

{
  "image_url": "https://raw.githubusercontent.com/facebookresearch/segment-anything-2/main/notebooks/images/truck.jpg",
  "prompt": "wheel",
  "point_prompts": [],
  "box_prompts": [],
  "apply_mask": true,
  "output_format": "png",
  "max_masks": 3
}

Output#

image Image

Primary segmented mask preview.

masks list<Image>* required

Segmented mask images.

metadata list<MaskMetadata>

Per-mask metadata including scores and boxes.

scores list<float>

Per-mask confidence scores when requested.

boxes list<list<float>>

Per-mask normalized bounding boxes [cx, cy, w, h] when requested.

{
  "masks": [
    {
      "url": "",
      "content_type": "image/png",
      "file_name": "z9RV14K95DvU.png",
      "file_size": 4404019,
      "width": 1024,
      "height": 1024
    }
  ],
  "metadata": [
    {}
  ]
}

Other types#

PointPrompt#

x integer

X Coordinate of the prompt

y integer

Y Coordinate of the prompt

label Enum

1 for foreground, 0 for background

Possible enum values: 0, 1

object_id integer

Optional object identifier. Prompts sharing an object id refine the same object.

frame_index integer

The frame index to interact with.

Image#

url string* required

The URL where the file can be downloaded from.

content_type string

The mime type of the file.

file_name string

The name of the file. It will be auto-generated if not provided.

file_size integer

The size of the file in bytes.

width integer

The width of the image in pixels.

height integer

The height of the image in pixels.

MaskMetadata#

index integer* required

Index of the mask inside the model output.

score float

Score for this mask.

box list<float>

Bounding box for the mask in normalized cxcywh coordinates.

SAM3DBodyPersonMetadata#

person_id integer* required

Index of the person in the scene

bbox list<float>* required

Bounding box [x_min, y_min, x_max, y_max]

focal_length float* required

Estimated focal length

pred_cam_t list<float>* required

Predicted camera translation [tx, ty, tz]

keypoints_2d list<list<float>>* required

2D keypoints [[x, y], ...] - 70 MHR body keypoints in image coordinates. See SAM3DBodyMetadata.keypoint_names for the ordered name of each index.

keypoints_3d list<list<float>>

3D keypoints [[x, y, z], ...] - 70 MHR body keypoints in camera space. Ordering matches keypoints_2d / SAM3DBodyMetadata.keypoint_names.

shape_params list<float>

MHR identity (β) shape parameters. Enables local canonical-pose mesh reconstruction when combined with the MHR model.

body_pose_params list<list<float>>

Per-joint body pose parameters (axis-angle form) from the MHR model.

hand_pose_params list<list<float>>

Per-joint hand pose parameters (axis-angle form) from the MHR model.

global_rot list<void>

Global root rotation produced by MHR. Shape matches the upstream tensor (axis-angle [3] or rotation matrix [3, 3]).

pred_global_rots list<void>

Per-joint global rotations (world-space), typically [N_joints, 3, 3] rotation matrices. Needed for inverse linear-blend skinning / un-posing clients.

scale_params list<float>

MHR scale parameters (isotropic or per-axis).

expr_params list<float>

MHR facial-expression parameters.

pred_joint_coords list<list<float>>

Skeleton joint positions in world space [[x, y, z], ...]. One row per MHR joint.

mhr_model_params list<void>

Packed MHR parameter vector (concatenated shape/pose/expression/scale). Shape is forwarded as-is from the upstream model.

pred_pose_raw list<void>

Raw pose transforms produced by the MHR decoder (pre-FK), forwarded as-is from the upstream model.

BoxPrompt#

x_min integer

X Min Coordinate of the box

y_min integer

Y Min Coordinate of the box

x_max integer

X Max Coordinate of the box

y_max integer

Y Max Coordinate of the box

object_id integer

Optional object identifier. Boxes sharing an object id refine the same object.

frame_index integer

The frame index to interact with.

SAM3DBodyMetadata#

num_people integer* required

Number of people detected

people list<SAM3DBodyPersonMetadata>* required

Per-person metadata

keypoint_names list<string>

Ordered names of the 70 MHR keypoints returned by this endpoint. Index i in this list corresponds to index i in every person's keypoints_2d and keypoints_3d arrays. Sourced from facebookresearch/sam-3d-body mhr70.py.

File#

url string* required

The URL where the file can be downloaded from.

content_type string

The mime type of the file.

file_name string

The name of the file. It will be auto-generated if not provided.

file_size integer

The size of the file in bytes.

SAM3DBodyAlignmentInfo#

person_id integer* required

Index of the person

scale_factor float* required

Scale factor applied for alignment

translation list<float>* required

Translation [tx, ty, tz]

focal_length float* required

Focal length used

target_points_count integer* required

Number of target points for alignment

cropped_vertices_count integer* required

Number of cropped vertices

BoxPromptBase#

x_min integer

X Min Coordinate of the box

y_min integer

Y Min Coordinate of the box

x_max integer

X Max Coordinate of the box

y_max integer

Y Max Coordinate of the box

object_id integer

Optional object identifier. Boxes sharing an object id refine the same object.

SAM3DObjectMetadata#

object_index integer* required

Index of the object in the scene

scale list<list<float>>

Scale factors [sx, sy, sz]

rotation list<list<float>>

Rotation quaternion [x, y, z, w]

translation list<list<float>>

Translation [tx, ty, tz]

camera_pose list<list<float>>

Camera pose matrix

PointPromptBase#

x integer

X Coordinate of the prompt

y integer

Y Coordinate of the prompt

label Enum

1 for foreground, 0 for background

Possible enum values: 0, 1

object_id integer

Optional object identifier. Prompts sharing an object id refine the same object.

fal-ai/sam-3/image

Table of contents

1. Calling the API

2. Authentication

3. Queue

4. Files

5. Schema

About

1. Calling the API#

Install the client#

Migrate to @fal-ai/client

Setup your API Key#

Submit a request#

Real-time via WebSockets#

2. Authentication#

API Key#

Protect your API Key

3. Queue#

Long-running requests

Submit a request#

Fetch request status#

Get the result#

4. Files#

Data URI (base64)#

Hosted files (URL)#

Uploading files#

Auto uploads

5. Schema#

Input#

Output#

Other types#

PointPrompt#

Image#

MaskMetadata#

SAM3DBodyPersonMetadata#

BoxPrompt#

SAM3DBodyMetadata#

File#

SAM3DBodyAlignmentInfo#

BoxPromptBase#

SAM3DObjectMetadata#

PointPromptBase#

Related Models