def __init__(self, do_cuda=True):
self.lock = RLock()
self.opts = rfnetOptions(do_cuda)
print(self.opts)
print('rfnet init')
#print(f"{gct()} model init")
print("model init")
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(
det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE, cfg.TRAIN.TOPK
)
device = torch.device("cuda")
model = model.to(device)
#resume = args.resume
#resume = "/home/cviss3/PycharmProjects/gensynth_dev_env/pyslam/thirdparty/rfnet/runs/10_24_09_25/model/e121_NN_0.480_NNT_0.655_NNDR_0.813_MeanMS_0.649.pth.tar"
resume = "/content/RFnetpyslam/thirdparty/rfnet/runs/10_24_09_25/model/e121_NN_0.480_NNT_0.655_NNDR_0.813_MeanMS_0.649.pth.tar"
print('==> Loading pre-trained network.')
checkpoint = torch.load(resume)
model.load_state_dict(checkpoint["state_dict"])
self.fe = model
print('==> Successfully loaded pre-trained network.')
self.device = device
self.pts = []
self.kps = []
self.des = []
self.img = []
self.heatmap = []
self.frame = None
self.frameFloat = None
self.keypoint_size = 20 # just a representative size for visualization and in order to convert extracted points to cv2.KeyPoint
print(f"{gct()} : Build the model")
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE,
cfg.TRAIN.TOPK)
if mgpu:
model = torch.nn.DataParallel(model)
model = model.to(device=device)
###############################################################################
# Load train data
###############################################################################
PPT = [cfg.PROJ.TRAIN_PPT, (cfg.PROJ.TRAIN_PPT + cfg.PROJ.EVAL_PPT)]
print(f"{gct()} : Loading traning data")
train_data = DataLoader(
HpatchDataset(
data_type="train",
PPT=PPT,
use_all=cfg.PROJ.TRAIN_ALL,
model_file = "./model/e121_NN_0.480_NNT_0.655_NNDR_0.813_MeanMS_0.649.pth.tar"
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE,
cfg.TRAIN.TOPK)
model = model.to(device)
checkpoint = torch.load(model_file)
model.load_state_dict(checkpoint["state_dict"])
model.eval()
def distance_matrix_vector(anchor, positive):
eps = 1e-8
FeatSimi_Mat = 2 - 2 * torch.mm(anchor, positive.t()) # [0, 4]
FeatSimi_Mat = FeatSimi_Mat.clamp(min=eps, max=4.0)
FeatSimi_Mat = torch.sqrt(FeatSimi_Mat) # euc [0, 2]
return FeatSimi_Mat
#输入模型参数
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE,
cfg.TRAIN.TOPK)
print(f"{gct()} : to device")
device = torch.device("cuda")
model = model.to(device)
resume = args.resume
print(f"{gct()} : in {resume}")
checkpoint = torch.load(resume)
model.load_state_dict(checkpoint["state_dict"])
###############################################################################
# detect and compute
###############################################################################
#split表示通过分隔符对字符串进行切割
img1_path, img2_path = args.imgpath.split("@")
kp1, des1, img1 = model.detectAndCompute(img1_path, device, (240, 320))
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
args.k,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(
det,
des,
cfg.LOSS.SCORE,
cfg.LOSS.PAIR,
cfg.PATCH.SIZE,
args.k,
)
model = model.to(device)
checkpoint = torch.load(model_file)
model.load_state_dict(checkpoint["state_dict"])
model.eval()
random.seed(cfg.PROJ.SEED)
torch.manual_seed(cfg.PROJ.SEED)
np.random.seed(cfg.PROJ.SEED)
root_dir = '/home/wang/workspace/RFSLAM_offline/RFNET/data/'
csv_file = None
seq = None
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
1024,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(
det,
des,
cfg.LOSS.SCORE,
cfg.LOSS.PAIR,
cfg.PATCH.SIZE,
512,
)
model = model.to(device)
flop, para = profile(det, (torch.randn((1, 1, 640, 480), device=device), ))
'''
flop
6829056000.0
para
Out[4]: 21890.0
'''
flop2, para2 = profile(des, (torch.randn((1000, 1, 32, 32), device=device), ))
'''
flop2
model_file = "/home/wang/workspace/CVToolbox/benchmark/evaluationMMA/model/e121_NN_0.480_NNT_0.655_NNDR_0.813_MeanMS_0.649.pth.tar"
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE,
cfg.TRAIN.TOPK)
model = model.to(device)
checkpoint = torch.load(model_file)
model.load_state_dict(checkpoint["state_dict"])
model.eval()
img1_path = "./material/4_1.png"
img2_path = "./material/4_2.png"
img1 = cv2.imread(img1_path)
img2 = cv2.imread(img2_path)
width = img1.shape[1]
kp1, des1, _ = model.detectAndCompute(img1_path, device,
(img1.shape[0], img1.shape[1]))
kp2, des2, _ = model.detectAndCompute(img2_path, device,
(img2.shape[0], img2.shape[1]))
print(f"{gct()} : model init")
det = RFDetSO(
cfg.TRAIN.score_com_strength,
cfg.TRAIN.scale_com_strength,
cfg.TRAIN.NMS_THRESH,
cfg.TRAIN.NMS_KSIZE,
cfg.TRAIN.TOPK,
cfg.MODEL.GAUSSIAN_KSIZE,
cfg.MODEL.GAUSSIAN_SIGMA,
cfg.MODEL.KSIZE,
cfg.MODEL.padding,
cfg.MODEL.dilation,
cfg.MODEL.scale_list,
)
des = HardNetNeiMask(cfg.HARDNET.MARGIN, cfg.MODEL.COO_THRSH)
model = RFNetSO(det, des, cfg.LOSS.SCORE, cfg.LOSS.PAIR, cfg.PATCH.SIZE,
cfg.TRAIN.TOPK)
print(f"{gct()} : to device")
device = torch.device("cuda")
model = model.to(device)
resume = '/home/wang/workspace/Faster-net2/runs/10_24_09_25/model/e121_NN_0.480_NNT_0.655_NNDR_0.813_MeanMS_0.649.pth.tar'
print(f"{gct()} : in {resume}")
checkpoint = torch.load(resume)
model.load_state_dict(checkpoint["state_dict"])
def to_cv2_kp(kp, scale, angle):
# kp is like [batch_idx, y, x, channel]
return cv2.KeyPoint(kp[2], kp[1], scale, angle)