RK3588 InsightFace人脸识别移植及精度测试全解析
一、背景介绍
在边缘计算领域，RK3588作为一款高性能AIoT芯片，具备6TOPS的NPU算力，非常适合部署人脸识别等计算机视觉应用。本文将详细讲解如何将先进的InsightFace人脸识别模型移植到RK3588平台，并进行精度验证。
为什么选择InsightFace？
InsightFace是当前最先进的开源人脸识别项目之一，其核心是基于深度卷积神经网络的特征提取模型。与传统方法相比，它能生成具有更强区分度的128维人脸特征向量（embedding），即使在不同光照、角度和遮挡条件下也能保持高识别精度。
二、RK3588测试数据
- 模型Shape:1x3x112x112
- 精度类型:INT8
- 推理耗时:14.23 ms
- 平均准确率: 99.43% 官方精度: 99.52%
- 最佳阈值: 0.2740
- 左图：正负样本相似度分布
理想情况：同人相似度>0.4，不同人<0.2- 右图：ROC曲线
曲线越靠近左上角，模型性能越好（AUC>0.99为优秀）
三、环境搭建
四、模型获取与转换
4.1 获取源代码
git clone https://github.com/TreB1eN/InsightFace_Pytorch.git
cd InsightFace_Pytorch
4.2 下载预训练模型(model_ir_se50.pth)
4.2 下载LFW人脸对齐后的测试集并解压(lfw112.tar.gz)
4.3 PyTorch转ONNX
cat> torch2onnx.py <<-'EOF'
from model import Backbone, Arcface, MobileFaceNet, Am_softmax, l2_norm
import torch
model = Backbone(50,0.6, 'ir_se').eval()
model.load_state_dict(torch.load("model_ir_se50.pth",map_location=torch.device('cpu')))
dummy_input = torch.randn(1, 3, 112, 112)
torch.onnx._export(model, dummy_input.to("cpu"),"face_rec.onnx",
export_params=True, verbose=False,opset_version=11)
EOF
python3.10 torch2onnx.py
为什么需要ONNX中间格式？
ONNX（Open Neural Network Exchange）是通用的模型交换格式：
- 解决框架差异（PyTorch/TensorFlow等）
- 标准化模型结构，便于后续优化
- 支持跨平台部署（RKNN/TensorRT等）
4.4 ONNX转RKNN量化模型
cat> onnx2rknn.py <<-'EOF'
import os
import numpy as np
import cv2
from rknn.api import RKNN
from math import exp
ONNX_MODEL = 'face_rec.onnx'
RKNN_MODEL = 'face_rec.rknn'
is_quant=1
rknn = RKNN(verbose=True)
rknn.config(mean_values=[[127.5,127.5,127.5]],
std_values=[[127.5,127.5,127.5]], target_platform='rk3588')
ret = rknn.load_onnx(model=ONNX_MODEL)
if ret != 0:
print('Load model failed!')
exit(ret)
ret = rknn.build(do_quantization=is_quant, dataset='./dataset.txt',auto_hybrid=True)
if ret != 0:
print('Build model failed!')
exit(ret)
ret = rknn.export_rknn(RKNN_MODEL)
if ret != 0:
print('Export rknn model failed!')
exit(ret)
rknn.release()
EOF
find lfw112/ -name "*.jpg" | head -n 32 > dataset.txt
python3.10 onnx2rknn.py
量化原理：
量化将FP32权重转换为INT8格式：
原始范围：[-2.3, 5.1] → 映射到：0~255
缩放系数 = 255 / (5.1 - (-2.3)) ≈ 34.5
优势：
- 模型体积缩小75%（4字节→1字节）
- 内存带宽需求降低
- NPU运算速度提升3-5倍
注意事项：
- 使用真实数据（
dataset.txt中的图片）校准量化参数 - 避免使用纯色/无意义图片导致量化误差
五、精度验证
5.1 LFW数据集说明
Labeled Faces in the Wild（LFW）是权威人脸识别测试集：
- 包含5749个人的13,233张面部图像
- 提供6000对正负样本（3000对相同/3000对不同人）
- 测试结果可横向对比学术论文
5.2 精度测试
cat> lfw_evaluation.py <<-'EOF'
import os
import time
import matplotlib
matplotlib.use('agg')
import numpy as np
from PIL import Image
from sklearn.model_selection import KFold
from sklearn.metrics import accuracy_score
import matplotlib.pyplot as plt
from tqdm import tqdm
import cv2
import time
import numpy as np
from rknnlite.api import RKNNLite
class FaceRec(object):
def __init__(self):
RKNN_MODEL = 'face_rec.rknn'
self.rknn_lite = RKNNLite()
ret = self.rknn_lite.load_rknn(RKNN_MODEL)
if ret != 0:
print('Load RKNN model failed')
exit(ret)
ret = self.rknn_lite.init_runtime(core_mask=RKNNLite.NPU_CORE_AUTO)
if ret != 0:
print('Init runtime environment failed')
exit(ret)
self.model_width=112
self.model_height=112
self.count=0
def get_embedding(self,orig_img):
img_rgb = cv2.cvtColor(orig_img, cv2.COLOR_BGR2RGB)
blob = img_rgb.astype(np.uint8)
blob = blob[np.newaxis]
t0=time.time()
outputs = self.rknn_lite.inference(inputs=[blob],data_format=['nhwc'])
t1=time.time()
self.count+=1
if self.count==10:
print(f"Infer Time:{(t1-t0)*1000:.4f} ms")
return outputs[0][0]
# LFW数据集路径配置
LFW_DIR = "lfw112/aligned"
PAIRS_FILE = "lfw112/pairs.txt"
def load_lfw_pairs(pairs_filepath):
"""
加载LFW pairs.txt文件
返回格式: [(path1, path2, same_flag), ...]
"""
pairs = []
with open(pairs_filepath, 'r') as f:
for line in f.readlines()[1:]: # 跳过标题行
pair = line.strip().split()
if len(pair) == 3:
# 相同人
name, id1, id2 = pair
path1 = os.path.join(LFW_DIR, name, f"{name}_{id1.zfill(4)}.jpg")
path2 = os.path.join(LFW_DIR, name, f"{name}_{id2.zfill(4)}.jpg")
pairs.append((path1, path2, 1))
elif len(pair) == 4:
# 不同人
name1, id1, name2, id2 = pair
path1 = os.path.join(LFW_DIR, name1, f"{name1}_{id1.zfill(4)}.jpg")
path2 = os.path.join(LFW_DIR, name2, f"{name2}_{id2.zfill(4)}.jpg")
pairs.append((path1, path2, 0))
return pairs
def load_and_preprocess_image(image_path):
"""
加载并预处理图像
返回PIL.Image对象 (RGB格式)
"""
img = cv2.imread(image_path)
return img
def compute_cosine_similarity(emb1, emb2):
"""计算两个特征向量之间的余弦相似度"""
return np.dot(emb1, emb2) / (np.linalg.norm(emb1) * np.linalg.norm(emb2))
def evaluate_lfw(model, pairs, batch_size=32):
"""
在LFW数据集上评估模型
参数:
model: 人脸识别模型实例
pairs: LFW对列表 [(path1, path2, same_flag), ...]
batch_size: 批处理大小
返回:
accuracy: 平均准确率
thresholds: 最佳阈值列表
similarities: 所有对的相似度分数
labels: 所有对的真实标签
"""
similarities = []
labels = []
thresholds = []
# 进度条
pbar = tqdm(total=len(pairs), desc="Processing pairs")
# 分批处理
for i in range(0, len(pairs), batch_size):
batch_pairs = pairs[i:i+batch_size]
batch_embs1 = []
batch_embs2 = []
# 加载并提取特征
for path1, path2, label in batch_pairs:
try:
img1 = load_and_preprocess_image(path1)
img2 = load_and_preprocess_image(path2)
emb1 = model.get_embedding(img1)
emb2 = model.get_embedding(img2)
batch_embs1.append(emb1)
batch_embs2.append(emb2)
labels.append(label)
except Exception as e:
print(f"Error processing {path1} or {path2}: {str(e)}")
continue
# 计算相似度
for emb1, emb2 in zip(batch_embs1, batch_embs2):
sim = compute_cosine_similarity(emb1, emb2)
similarities.append(sim)
pbar.update(len(batch_pairs))
pbar.close()
# 转换为numpy数组
similarities = np.array(similarities)
labels = np.array(labels)
# 10折交叉验证寻找最佳阈值
kf = KFold(n_splits=10, shuffle=True)
accuracies = []
for train_idx, test_idx in kf.split(similarities):
train_sims = similarities[train_idx]
train_labels = labels[train_idx]
# 寻找最佳阈值
best_threshold = 0
best_accuracy = 0
for threshold in np.arange(-1.0, 1.0, 0.01):
preds = (train_sims > threshold).astype(int)
acc = accuracy_score(train_labels, preds)
if acc > best_accuracy:
best_accuracy = acc
best_threshold = threshold
# 在测试集上评估
test_sims = similarities[test_idx]
test_labels = labels[test_idx]
test_preds = (test_sims > best_threshold).astype(int)
test_acc = accuracy_score(test_labels, test_preds)
accuracies.append(test_acc)
thresholds.append(best_threshold)
# 计算平均准确率
mean_accuracy = np.mean(accuracies) * 100
mean_threshold = np.mean(thresholds)
return mean_accuracy, mean_threshold, similarities, labels
def plot_results(similarities, labels, threshold):
"""绘制相似度分布和ROC曲线"""
# 相似度分布图
plt.figure(figsize=(15, 5))
plt.subplot(1, 2, 1)
same_sims = similarities[labels == 1]
diff_sims = similarities[labels == 0]
plt.hist(same_sims, bins=50, alpha=0.7, label='Same Person', color='green')
plt.hist(diff_sims, bins=50, alpha=0.7, label='Different Person', color='red')
plt.axvline(x=threshold, color='blue', linestyle='--', label=f'Threshold: {threshold:.2f}')
plt.xlabel('Cosine Similarity')
plt.ylabel('Count')
plt.title('Similarity Distribution')
plt.legend()
# ROC曲线
plt.subplot(1, 2, 2)
tpr_list = []
fpr_list = []
thresholds = np.arange(-1.0, 1.0, 0.01)
for thresh in thresholds:
preds = (similarities > thresh).astype(int)
tp = np.sum((preds == 1) & (labels == 1))
fp = np.sum((preds == 1) & (labels == 0))
tn = np.sum((preds == 0) & (labels == 0))
fn = np.sum((preds == 0) & (labels == 1))
tpr = tp / (tp + fn) if (tp + fn) > 0 else 0
fpr = fp / (fp + tn) if (fp + tn) > 0 else 0
tpr_list.append(tpr)
fpr_list.append(fpr)
plt.plot(fpr_list, tpr_list)
plt.plot([0, 1], [0, 1], 'k--') # 随机猜测线
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('ROC Curve')
plt.grid(True)
plt.tight_layout()
plt.savefig('lfw_results.png', dpi=300)
plt.show()
if __name__ == "__main__":
# 1. 初始化模型
model = FaceRec()
# 2. 加载LFW数据对
print("加载LFW数据集...")
pairs = load_lfw_pairs(PAIRS_FILE)
print(f"加载完成! 共有 {len(pairs)} 对图像")
# 3. 评估模型
print("开始评估模型...")
start_time = time.time()
accuracy, threshold, similarities, labels = evaluate_lfw(model, pairs)
elapsed = time.time() - start_time
# 4. 打印结果
print("\n" + "="*50)
print(f"LFW测试结果:")
print(f"平均准确率: {accuracy:.2f}%")
print(f"最佳阈值: {threshold:.4f}")
print(f"总耗时: {elapsed:.2f}秒")
print("="*50)
# 5. 可视化结果
print("生成结果可视化...")
plot_results(similarities, labels, threshold)
print("结果图已保存为 'lfw_results.png'")
EOF
pip3.10 install scikit-learn -i https://mirrors.aliyun.com/pypi/simple/
python3.10 lfw_evaluation.py
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
