def __init__(self, motionType):
self.motionType = motionType.replace(" ", "").lower()
self.status = 0
self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
self.lb = joblib.load("models/customized/" + self.motionType + "/outputs/lb_" + self.motionType + ".pkl")
self.model = cnn_models.CustomCNN(self.lb)
self.model.load_state_dict(
torch.load("models/customized/" + self.motionType + "/outputs/" + self.motionType + ".pth"))
self.aug = albumentations.Compose([albumentations.Resize(224, 224),])
self.model = self.model.eval().to(self.device)
self.labels = []
self.isPressed = False
def __init__(self, motionType):
self.motionType = motionType
self.prepare_data()
self.lr = 1e-3
self.batch_size = 32
self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
self.lb = joblib.load('models/customized/' + self.motionType + '/outputs/lb_' + self.motionType + '.pkl')
self.model = cnn_models.CustomCNN(self.lb).to(self.device)
self.epochs = 75
self.epoch = 0
df = pd.read_csv('models/customized/' + self.motionType + '/input/data_' + self.motionType + '.csv')
X = df.image_path.values
y = df.target.values
(xtrain, xtest, ytrain, ytest) = train_test_split(X, y, test_size=0.10, random_state=42)
print(f"Training instances: {len(xtrain)}")
print(f"Validation instances: {len(xtest)}")
self.train_data = ImageDataset(xtrain, ytrain, tfms=1)
self.test_data = ImageDataset(xtest, ytest, tfms=0)
self.trainloader = DataLoader(self.train_data, batch_size=self.batch_size, shuffle=True)
self.testloader = DataLoader(self.test_data, batch_size=self.batch_size, shuffle=False)
total_params = sum(p.numel() for p in self.model.parameters())
print(f"{total_params:,} total parameters.")
total_trainable_params = sum(
p.numel() for p in self.model.parameters() if p.requires_grad)
print(f"{total_trainable_params:,} training parameters.")
self.optimizer = optim.Adam(self.model.parameters(), lr=self.lr)
self.criterion = nn.CrossEntropyLoss()
self.scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(
self.optimizer,
mode='min',
patience=5,
factor=0.5,
min_lr=1e-6,
verbose=True
)
image = np.transpose(image, (2, 0, 1)).astype(np.float32)
label = self.y[i]
return torch.tensor(image,
dtype=torch.float), torch.tensor(label,
dtype=torch.long)
train_data = ASLImageDataset(xtrain, ytrain)
test_data = ASLImageDataset(xtest, ytest)
# dataloaders
trainloader = DataLoader(train_data, batch_size=32, shuffle=True)
testloader = DataLoader(test_data, batch_size=32, shuffle=False)
# model = models.MobineNetV2(pretrained=True, requires_grad=False)
model = cnn_models.CustomCNN().to(device)
print(model)
# total parameters and trainable parameters
total_params = sum(p.numel() for p in model.parameters())
print(f"{total_params:,} total parameters.")
total_trainable_params = sum(p.numel() for p in model.parameters()
if p.requires_grad)
print(f"{total_trainable_params:,} training parameters.")
# optimizer
optimizer = optim.Adam(model.parameters(), lr=0.001)
# loss function
criterion = nn.CrossEntropyLoss()
# training function
USAGE:
python cam_test.py
'''
import torch
import joblib
import torch.nn as nn
import numpy as np
import cv2
import torch.nn.functional as F
import time
import cnn_models
# load label binarizer
lb = joblib.load(
'C:\\Users\\SARAH\\Desktop\\Multivariate Stats\\ASL\\output\\lb.pkl')
model = cnn_models.CustomCNN()
model.load_state_dict(
torch.load(
'C:\\Users\\SARAH\\Desktop\\Multivariate Stats\\ASL\\output\\model.pth'
))
print(model)
print('Model loaded')
# In[ ]:
def hand_area(img):
hand = img[100:324, 100:324]
hand = cv2.resize(hand, (224, 224))
return hand
'--label-bin',
required=True,
help="path to label binarizer")
ap.add_argument('-i', '--input', required=True, help='path to our input video')
ap.add_argument('-o',
'--output',
required=True,
type=str,
help='path to our output video')
args = vars(ap.parse_args())
# load the trained model and label binarizer from disk
print('Loading model and label binarizer...')
lb = joblib.load(args['label_bin'])
model = cnn_models.CustomCNN().cuda()
print('Model Loaded...')
model.load_state_dict(torch.load(args['model']))
print('Loaded model state_dict...')
aug = albumentations.Compose([
albumentations.Resize(224, 224),
])
cap = cv2.VideoCapture(args['input'])
if (cap.isOpened() == False):
print('Error while trying to read video. Plese check again...')
# get the frame width and height