Gesture-Control/utils.py at master · rishabhpatel9/Gesture-Control · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
import matplotlib.pyplot as plt
import torch
import numpy as np
from matplotlib import animation
from IPython.display import display, HTML
#plt.rcParams['animation.ffmpeg_path'] = '/home/fabio/anaconda3/bin/ffmpeg'
plt.rcParams['animation.ffmpeg_path'] = '/home/fabio/anaconda3/envs/ai/bin/ffmpeg'


def plot_batch_mp4(video_array, std=None, mean=None):
	for video in video_array:
		plot_movie_mp4(video, std=std, mean=mean)

def plot_labeled_batch_mp4(video_array, labels, std=None, mean=None):
	for video, label in zip(video_array, labels):
		plot_movie_mp4(video, label, std=std, mean=mean)


def plot_movie_mp4(image_array, text=None, std=None, mean=None):
	if std and mean:
		tensor_list = torch.tensor(image_array.permute(1,0,2,3))
		for t, m, s in zip(image_array, mean, std):
			t.mul_(s).add_(m)
		image_array = torch.clamp(tensor_list.permute(1,2,3,0), 0, 1).numpy()
	else:
		image_array = image_array.permute(0,2,3,1).numpy()
	dpi = 60.0
	xpixels, ypixels = image_array[0].shape[0], image_array[0].shape[1]
	fig = plt.figure(figsize=(ypixels/dpi, xpixels/dpi), dpi=dpi);
	im = plt.figimage(image_array[0]);

	def animate(i):
		im.set_array(image_array[i])
		return (im)

	anim = animation.FuncAnimation(fig, animate, frames=len(image_array))

	if text:
		prepend = '<p>'+ str(text) +'</p>'
		display(HTML(prepend + anim.to_html5_video()))
	else:
		display(HTML(anim.to_html5_video()))

def calculate_loss_and_accuracy(validation_loader, model, criterion, stop_at = 1200, print_every=99999):
	correct = 0
	total = 0
	steps = 0
	total_loss = 0
	sz = len(validation_loader)

	for images, labels in validation_loader:

		if total%print_every == 0 and total > 0:
			accuracy = 100 * correct / total
			print(accuracy)

		if total >= stop_at:
			break;
		if torch.cuda.is_available():
			images = images.cuda()
			labels = labels.cuda()

		# Forward pass only to get logits/output
		outputs = model(images)

		#Get Loss for validation data
		loss = criterion(outputs, labels)
		total_loss += loss.item()


		# Get predictions from the maximum value
		_, predicted = torch.max(outputs.data, 1)

		# Total number of labels
		total += labels.size(0)
		steps += 1

		correct += (predicted == labels).sum().item()

		del outputs, loss, _, predicted

	accuracy = 100 * correct / total
	return total_loss/steps, accuracy

def calculate_loss_and_tops(validation_loader, model, criterion, stop_at = 1200, print_every=99999):
	correct, correct3, correct5 = 0, 0, 0
	total = 0
	steps = 0
	total_loss = 0
	sz = len(validation_loader)
	model.eval()

	for images, labels in validation_loader:

		if total%print_every == 0 and total > 0:
			top1 = 100 * correct / total
			top3 = 100 * correct3 / total
			top5 = 100 * correct5 / total
			print(top1, top3, top5)

		if total >= stop_at:
			break;
		if torch.cuda.is_available():
			images = images.cuda()
			labels = labels.cuda()

		# Forward pass only to get logits/output
		outputs = model(images)

		#Get Loss for validation data
		loss = criterion(outputs, labels)
		total_loss += loss.item()


		# Get predictions from the maximum value
		_, predicted = torch.max(outputs.data, 1)
		_, predicted3 = torch.topk(outputs.data, 3)
		_, predicted5 = torch.topk(outputs.data, 5)

		top3 = [l in p for (l,p) in zip(labels, predicted3)]
		top5 = [l in p for (l,p) in zip(labels, predicted5)]

		# Total number of labels
		total += labels.size(0)
		steps += 1

		correct += (predicted == labels).sum().item()
		correct3 += np.sum(top3)
		correct5 += np.sum(top5)


		del outputs, loss, _, predicted

		top1 = 100 * correct / total
		top3 = 100 * correct3 / total
		top5 = 100 * correct5 / total

	return total_loss/steps, top1, top3, top5