Learn data science and machine learning by building real-world projects on Jovian
Sign up to execute sign-detection-4b69a and 160,000+ data science projects. Build your own projects and share them online!
Created 9 months ago
Sign Language Detection - An Image Classification Project
About
The data set is a collection of images of alphabets from the American Sign Language, separated in 29 folders which represent the various classes.
Content
The training data set contains 87,000 images which are 200x200 pixels. There are 29 classes, of which 26 are for the letters A-Z and 3 classes for SPACE, DELETE and NOTHING. These 3 classes are very helpful in real-time applications, and classification. The test data set contains a mere 28 images, to encourage the use of real-world test images.
Acknowledgements
- How to use Kaggle Datasets in Google Colab - https://medium.com/analytics-vidhya/how-to-fetch-kaggle-datasets-into-google-colab-ea682569851a
- Reference ipynb notebook - https://colab.research.google.com/drive/1c-U-fCmpcI0TAIds4_9IatzxMkcYnY8d#scrollTo=xfPP3dVf3KUg
Importing important libraries and project creation.
In [1]:
import os
import torch
import torchvision
import tarfile
import torch.nn as nn
import numpy as np
import torch.nn.functional as F
from torchvision.datasets.utils import download_url
from torchvision.datasets import ImageFolder
from torch.utils.data import DataLoader
import torchvision.transforms as tt
from torch.utils.data import random_split
from torchvision.utils import make_grid
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline
matplotlib.rcParams['figure.facecolor'] = '#ffffff'In [2]:
project_name='sign-detection'Retrieving sign language dataset from Kaggle datasets.
In [3]:
from google.colab import drive
drive.mount('/content/gdrive')Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount("/content/gdrive", force_remount=True).
In [4]:
import os
os.environ['KAGGLE_CONFIG_DIR'] = "/content/gdrive/My Drive/Kaggle"In [5]:
%cd /content/gdrive/My Drive/Kaggle/content/gdrive/My Drive/Kaggle
In [6]:
!kaggle datasets download -d grassknoted/asl-alphabetasl-alphabet.zip: Skipping, found more recently modified local copy (use --force to force download)
In [7]:
!unzip \*.zip && rm *.zipArchive: asl-alphabet.zip
replace asl_alphabet_test/asl_alphabet_test/A_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/B_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/C_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/D_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/E_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/F_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/G_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/H_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/I_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
n
replace asl_alphabet_test/asl_alphabet_test/J_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: replace asl_alphabet_test/asl_alphabet_test/K_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/L_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
n
replace asl_alphabet_test/asl_alphabet_test/M_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: replace asl_alphabet_test/asl_alphabet_test/N_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/O_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/P_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/Q_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/R_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/S_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/T_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/U_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/V_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
n
replace asl_alphabet_test/asl_alphabet_test/W_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: replace asl_alphabet_test/asl_alphabet_test/X_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/Y_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/Z_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/nothing_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_test/asl_alphabet_test/space_test.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A10.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A100.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1000.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1001.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1002.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1003.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
n
replace asl_alphabet_train/asl_alphabet_train/A/A1004.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: replace asl_alphabet_train/asl_alphabet_train/A/A1005.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1006.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1007.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1008.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1009.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A101.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1010.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1011.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: n
replace asl_alphabet_train/asl_alphabet_train/A/A1012.jpg? [y]es, [n]o, [A]ll, [N]one, [r]ename: In [8]:
pwdOut[8]:
'/content/gdrive/My Drive/Kaggle'In [9]:
cd ../content/gdrive/My Drive
In [10]:
cd ../content/gdrive
In [11]:
cd ../content
In [12]:
cd ../
The above section is only to be used when the dataset is being retrived for the first time.
Preparing ASL Sign Language Dataset
In [13]:
from google.colab import drive
drive.mount('/content/gdrive')
os.environ['KAGGLE_CONFIG_DIR'] = "/content/gdrive/My Drive/Kaggle"Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount("/content/gdrive", force_remount=True).
In [14]:
data_dir = '/content/gdrive/MyDrive/Kaggle'
print(os.listdir(data_dir))
classes = os.listdir(data_dir + "/asl_alphabet_train/asl_alphabet_train")
print(classes)
['kaggle.json', 'asl_alphabet_test', 'asl_alphabet_train', 'flowers', 'asl-alphabet.zip']
['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 'del', 'nothing', 'space']
In [15]:
data_tfms = tt.Compose([tt.Resize((64,64)),tt.CenterCrop((64,64)),tt.ToTensor()])In [16]:
data_ds = ImageFolder(data_dir+'/asl_alphabet_train/asl_alphabet_train', data_tfms)
len(data_ds)Out[16]:
87000In [17]:
train_ds, valid_ds = torch.utils.data.random_split(data_ds, [78300,8700])Above I have not used the already present asl_alphabet_test folder since it has very less data for it to qualify being a proper validity check. I have thus randomly distributed the train data(new_train_data = 90%,new_test_data=10%) itself because it is very large.
In [18]:
img, label = train_ds[2500]
print(img.shape, label)
torch.Size([3, 64, 64]) 27
In [19]:
def show_example(img,label):
print('Label: ', classes[label], '('+str(label)+')')
plt.imshow(img.permute(1, 2, 0))
show_example(*train_ds[31419])Label: J (9)
In [20]:
batch_size = 256In [21]:
# PyTorch data loaders
train_dl = DataLoader(train_ds, batch_size, shuffle=True, num_workers=3, pin_memory=True)
valid_dl = DataLoader(valid_ds, batch_size*2, num_workers=3, pin_memory=True)In [22]:
from torchvision.utils import make_grid
def show_batch(dl):
for img, label in dl:
fig, ax = plt.subplots(figsize=(16, 12))
ax.set_xticks([]); ax.set_yticks([])
ax.imshow(make_grid(img, nrow = 16).permute(1,2,0))
breakIn [23]:
show_batch(train_dl)