Deteção de caixas limitadoras

Neste lançamento experimental, estamos a disponibilizar aos programadores uma ferramenta poderosa para a deteção e a localização de objetos em imagens e vídeos. Ao identificar e delinear com precisão objetos com caixas delimitadoras, os programadores podem desbloquear uma vasta gama de aplicações e melhorar a inteligência dos respetivos projetos.

Principais vantagens:

  • Simples: integre facilmente capacidades de deteção de objetos nas suas aplicações, independentemente da sua experiência em visão computacional.
  • Personalizável: produza caixas delimitadoras com base em instruções personalizadas (por exemplo, "Quero ver caixas delimitadoras de todos os objetos verdes nesta imagem"), sem ter de preparar um modelo personalizado.

Detalhes técnicos:

  • Entrada: o seu comando e as imagens ou os frames de vídeo associados.
  • Saída: caixas delimitadoras no formato [y_min, x_min, y_max, x_max]. O canto superior esquerdo é a origem. O eixo x move-se na horizontal e o eixo y move-se na vertical. Os valores das coordenadas são normalizados para 0 a 1000 para cada imagem.
  • Visualização: os utilizadores do AI Studio veem caixas delimitadoras representadas na IU. Os utilizadores da Vertex AI devem visualizar as respetivas caixas delimitadoras através de código de visualização personalizado.

Python

Instalação

pip install --upgrade google-genai

Para saber mais, consulte a documentação de referência do SDK.

Defina variáveis de ambiente para usar o SDK de IA generativa com o Vertex AI: 

# Replace the `GOOGLE_CLOUD_PROJECT` and `GOOGLE_CLOUD_LOCATION` values
# with appropriate values for your project.
export GOOGLE_CLOUD_PROJECT=GOOGLE_CLOUD_PROJECT
export GOOGLE_CLOUD_LOCATION=global
export GOOGLE_GENAI_USE_VERTEXAI=True
 

import requests
from google import genai
from google.genai.types import (
    GenerateContentConfig,
    HarmBlockThreshold,
    HarmCategory,
    HttpOptions,
    Part,
    SafetySetting,
)
from PIL import Image, ImageColor, ImageDraw
from pydantic import BaseModel

# Helper class to represent a bounding box
class BoundingBox(BaseModel):
    """
    Represents a bounding box with its 2D coordinates and associated label.

    Attributes:
        box_2d (list[int]): A list of integers representing the 2D coordinates of the bounding box,
                            typically in the format [y_min, x_min, y_max, x_max].
        label (str): A string representing the label or class associated with the object within the bounding box.
    """

    box_2d: list[int]
    label: str

# Helper function to plot bounding boxes on an image
def plot_bounding_boxes(image_uri: str, bounding_boxes: list[BoundingBox]) -> None:
    """
    Plots bounding boxes on an image with labels, using PIL and normalized coordinates.

    Args:
        image_uri: The URI of the image file.
        bounding_boxes: A list of BoundingBox objects. Each box's coordinates are in
                        normalized [y_min, x_min, y_max, x_max] format.
    """
    with Image.open(requests.get(image_uri, stream=True, timeout=10).raw) as im:
        width, height = im.size
        draw = ImageDraw.Draw(im)

        colors = list(ImageColor.colormap.keys())

        for i, bbox in enumerate(bounding_boxes):
            # Scale normalized coordinates to image dimensions
            abs_y_min = int(bbox.box_2d[0] / 1000 * height)
            abs_x_min = int(bbox.box_2d[1] / 1000 * width)
            abs_y_max = int(bbox.box_2d[2] / 1000 * height)
            abs_x_max = int(bbox.box_2d[3] / 1000 * width)

            color = colors[i % len(colors)]

            # Draw the rectangle using the correct (x, y) pairs
            draw.rectangle(
                ((abs_x_min, abs_y_min), (abs_x_max, abs_y_max)),
                outline=color,
                width=4,
            )
            if bbox.label:
                # Position the text at the top-left corner of the box
                draw.text((abs_x_min + 8, abs_y_min + 6), bbox.label, fill=color)

        im.show()

client = genai.Client(http_options=HttpOptions(api_version="v1"))

config = GenerateContentConfig(
    system_instruction="""
    Return bounding boxes as an array with labels.
    Never return masks. Limit to 25 objects.
    If an object is present multiple times, give each object a unique label
    according to its distinct characteristics (colors, size, position, etc..).
    """,
    temperature=0.5,
    safety_settings=[
        SafetySetting(
            category=HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT,
            threshold=HarmBlockThreshold.BLOCK_ONLY_HIGH,
        ),
    ],
    response_mime_type="application/json",
    response_schema=list[BoundingBox],
)

image_uri = "https://storage.googleapis.com/generativeai-downloads/images/socks.jpg"

response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents=[
        Part.from_uri(
            file_uri=image_uri,
            mime_type="image/jpeg",
        ),
        "Output the positions of the socks with a face. Label according to position in the image.",
    ],
    config=config,
)
print(response.text)
plot_bounding_boxes(image_uri, response.parsed)

# Example response:
# [
#     {"box_2d": [6, 246, 386, 526], "label": "top-left light blue sock with cat face"},
#     {"box_2d": [234, 649, 650, 863], "label": "top-right light blue sock with cat face"},
# ]

Go

Saiba como instalar ou atualizar o Go.

Para saber mais, consulte a documentação de referência do SDK.

Defina variáveis de ambiente para usar o SDK de IA generativa com o Vertex AI: 

# Replace the `GOOGLE_CLOUD_PROJECT` and `GOOGLE_CLOUD_LOCATION` values
# with appropriate values for your project.
export GOOGLE_CLOUD_PROJECT=GOOGLE_CLOUD_PROJECT
export GOOGLE_CLOUD_LOCATION=global
export GOOGLE_GENAI_USE_VERTEXAI=True
 

import (
	"context"
	"encoding/json"
	"fmt"
	"image"
	"image/color"
	"image/draw"
	"image/jpeg"
	"io"
	"net/http"

	"google.golang.org/genai"
)

// BoundingBox represents a bounding box with coordinates and label.
type BoundingBox struct {
	Box2D []int  `json:"box_2d"`
	Label string `json:"label"`
}

// plotBoundingBoxes downloads the image and overlays bounding boxes.
func plotBoundingBoxes(imageURI string, boundingBoxes []BoundingBox) error {
	resp, err := http.Get(imageURI)
	if err != nil {
		return fmt.Errorf("failed to download image: %w", err)
	}
	defer resp.Body.Close()

	img, err := jpeg.Decode(resp.Body)
	if err != nil {
		return fmt.Errorf("failed to decode image: %w", err)
	}

	bounds := img.Bounds()
	rgba := image.NewRGBA(bounds)
	draw.Draw(rgba, bounds, img, bounds.Min, draw.Src)

	// Simple red color for bounding boxes
	red := color.RGBA{255, 0, 0, 255}

	for _, bbox := range boundingBoxes {
		// scale normalized coordinates [0–1000] to absolute pixels
		yMin := bbox.Box2D[0] * bounds.Dy() / 1000
		xMin := bbox.Box2D[1] * bounds.Dx() / 1000
		yMax := bbox.Box2D[2] * bounds.Dy() / 1000
		xMax := bbox.Box2D[3] * bounds.Dx() / 1000

		// draw rectangle border
		for x := xMin; x <= xMax; x++ {
			rgba.Set(x, yMin, red)
			rgba.Set(x, yMax, red)
		}
		for y := yMin; y <= yMax; y++ {
			rgba.Set(xMin, y, red)
			rgba.Set(xMax, y, red)
		}
	}

	return nil
}

func generateBoundingBoxesWithText(w io.Writer) error {
	ctx := context.Background()

	client, err := genai.NewClient(ctx, &genai.ClientConfig{
		HTTPOptions: genai.HTTPOptions{APIVersion: "v1"},
	})
	if err != nil {
		return fmt.Errorf("failed to create genai client: %w", err)
	}

	imageURI := "https://storage.googleapis.com/generativeai-downloads/images/socks.jpg"

	// Schema definition for []BoundingBox
	schema := &genai.Schema{
		Type: genai.TypeArray,
		Items: &genai.Schema{
			Type: genai.TypeObject,
			Properties: map[string]*genai.Schema{
				"box_2d": {
					Type:  genai.TypeArray,
					Items: &genai.Schema{Type: genai.TypeInteger},
				},
				"label": {Type: genai.TypeString},
			},
			Required: []string{"box_2d", "label"},
		},
	}

	config := &genai.GenerateContentConfig{
		SystemInstruction: &genai.Content{
			Parts: []*genai.Part{{
				Text: "Return bounding boxes as an array with labels. Never return masks. Limit to 25 objects.",
			}},
		},
		Temperature:      float32Ptr(0.5),
		ResponseMIMEType: "application/json",
		ResponseSchema:   schema,
		SafetySettings: []*genai.SafetySetting{
			{
				Category:  genai.HarmCategoryDangerousContent,
				Threshold: genai.HarmBlockThresholdBlockOnlyHigh,
			},
		},
	}

	contents := []*genai.Content{
		{
			Role: "user",
			Parts: []*genai.Part{
				{
					FileData: &genai.FileData{
						FileURI:  imageURI,
						MIMEType: "image/jpeg",
					},
				},
				{Text: "Output the positions of the socks with a face. Label according to position in the image."},
			},
		},
	}

	resp, err := client.Models.GenerateContent(ctx, "gemini-2.5-flash", contents, config)
	if err != nil {
		return fmt.Errorf("failed to generate content: %w", err)
	}

	fmt.Fprintln(w, resp.Text())

	// Parse into []BoundingBox
	var boxes []BoundingBox
	if err := json.Unmarshal([]byte(resp.Text()), &boxes); err != nil {
		return fmt.Errorf("failed to parse bounding boxes: %w", err)
	}

	// Example response:
	//	Box: (962,113)-(2158,1631) Label: top left sock with face
	//	Box: (2656,721)-(3953,2976) Label: top right sock with face
	//...

	return plotBoundingBoxes(imageURI, boxes)
}

func float32Ptr(v float32) *float32 { return &v }