local_segment/sam1_demo.py

import warnings
warnings.filterwarnings("ignore")
import numpy as np
import torch
from segment_anything_hq import sam_model_registry, SamPredictor
import os,requests
from PIL import Image

def show_mask(mask):
    color = np.array([255, 255, 255, 1])
        
    h, w = mask.shape[-2:]
    mask = mask.astype(np.uint8)
    mask_image =  mask.reshape(h, w, 1) * color.reshape(1, 1, -1)
    # 将值域限制在 [0, 1]，然后转换为 [0, 255] 的 uint8 类型
    mask_image = np.clip(mask_image, 0, 1)  # 确保值在 [0, 1] 范围内
    mask_image_uint8 = (mask_image * 255).astype(np.uint8)
    return mask_image_uint8
def show_masks(masks, scores):
    for i, (mask, score) in enumerate(zip(masks, scores)):
        np_arr = show_mask(mask)
        arr_img = Image.fromarray(np_arr)
        arr_img.save(f"sam_1_ax_{i}.png")
sam_checkpoint = "./models/sam_hq_vit_h.pth"
model_type = "vit_h"
device = "cuda"
sam = sam_model_registry[model_type](checkpoint=sam_checkpoint)
sam.to(device=device)
predictor = SamPredictor(sam)
image_url = "https://ossimg.valimart.net/uploads/vali_ai/20260129/176968033473923.png"
raw_image = Image.open(requests.get(image_url, stream=True).raw).convert("RGB")
image_array = np.array(raw_image)
predictor.set_image(image_array)
input_point = np.array([[296, 543], [150, 543], [296, 683], [150, 683]])
input_label = np.array([1, 0, 1, 1])
hq_token_only = True 
masks, scores, logits = predictor.predict(
    point_coords=input_point,
    point_labels=input_label,
    multimask_output=True,
    hq_token_only=hq_token_only, 
)
mask_input = logits[np.argmax(scores), :, :]  # Choose the model's best mask
masks, scores, _ = predictor.predict(
    point_coords=input_point,
    point_labels=input_label,
    mask_input=mask_input[None, :, :],
    multimask_output=False,
     hq_token_only=hq_token_only, 
)
# masks = masks.squeeze(1).cpu().numpy()
show_masks(masks, scores)