Remote_sensing_image_classificatin-NPU

pip install -r requirements.txt -i https://pypi.tuna.tsinghua.edu.cn/simple

python3 -c "import torch; import torch_npu; print('NPU available:', torch.npu.is_available())"

# 安装 ModelScope
pip install modelscope -i https://pypi.tuna.tsinghua.edu.cn/simple

# 下载模型
from modelscope import snapshot_download
model_dir = snapshot_download('qzy3386401548/Remote_sensing_image_classificatin')

# 单架构推理
python inference.py --arch resnet34 --device cpu

# 全部架构推理
python inference.py --arch all --device cpu --output cpu_results.json

# 单架构推理
python inference.py --arch resnet34 --device npu

# 全部架构推理
python inference.py --arch all --device npu --output npu_results.json

python compare_cpu_npu.py --arch all --output compare_results.json

import torch
import numpy as np
from resnet34 import ResNet34

# 加载模型
model = ResNet34(num_classes=2)

# 移动到 NPU
model = model.to("npu:0")
model.eval()

# 准备输入 (模拟遥感图像)
dummy_input = torch.randn(1, 3, 224, 224).to("npu:0")

# 推理
with torch.no_grad():
    output = model(dummy_input)
    probs = torch.softmax(output, dim=-1)
    pred_class = torch.argmax(output, dim=-1)

print(f"预测类别: {pred_class.item()}, 概率: {probs.cpu().tolist()}")

import torch
from resnet34 import ResNet34

model = ResNet34(num_classes=2).to("npu:0")
model.eval()

# 批量输入
batch = torch.randn(16, 3, 224, 224).to("npu:0")

with torch.no_grad():
    outputs = model(batch)
    predictions = torch.argmax(outputs, dim=-1)

print(f"批量预测结果: {predictions.cpu().tolist()}")

# NPU 推理
python3 inference.py --device npu

# CPU 推理
python3 inference.py --device cpu