Add DCGAN app

Generator.py

@@ -0,0 +1,45 @@
+import torch
+import torch.nn as nn
+
+class Config:
+    workers = 2 # Number of workers for dataloader
+    batch_size = 128 # Batch size during training
+    image_size = 64
+    nc = 3 # Number of channels in the training images
+    nz = 100 # Size of z latent vector (i.e. size of generator input)
+    ngf = 64 # Size of feature maps in generator
+    ndf = 64 # Size of feature maps in discriminator
+    num_epochs = 15 # Number of training epochs
+    lr = 0.0002 # Learning rate for optimizers
+    beta1 = 0.5 # Beta1 hyperparam for Adam optimizers
+    ngpu = 1 # Number of GPUs available. Use 0 for CPU mode.
+
+class Generator(nn.Module):
+    def __init__(self, ngpu):
+        super(Generator, self).__init__()
+        self.ngpu = ngpu
+        self.network = nn.Sequential(
+            # input is Z, going into a convolution
+            nn.ConvTranspose2d(Config.nz, Config.ngf * 8, 4, 1, 0, bias=False),
+            nn.BatchNorm2d(Config.ngf * 8),
+            nn.ReLU(True),
+            # state size. (ngf*8) x 4 x 4
+            nn.ConvTranspose2d(Config.ngf * 8, Config.ngf * 4, 4, 2, 1, bias=False),
+            nn.BatchNorm2d(Config.ngf * 4),
+            nn.ReLU(True),
+            # state size. (ngf*4) x 8 x 8
+            nn.ConvTranspose2d(Config.ngf * 4, Config.ngf * 2, 4, 2, 1, bias=False),
+            nn.BatchNorm2d(Config.ngf * 2),
+            nn.ReLU(True),
+            # state size. (ngf*2) x 16 x 16
+            nn.ConvTranspose2d(Config.ngf * 2, Config.ngf, 4, 2, 1, bias=False),
+            nn.BatchNorm2d(Config.ngf),
+            nn.ReLU(True),
+            # state size. (ngf) x 32 x 32
+            nn.ConvTranspose2d(Config.ngf, Config.nc, 4, 2, 1, bias=False),
+            nn.Tanh()
+            # state size. (nc) x 64 x 64
+        )
+
+    def forward(self, xb):
+        return self.network(xb)

app.py

@@ -0,0 +1,37 @@
+import gradio as gr
+import torch
+from Generator import Generator
+from torchvision.utils import save_image
+
+
+generator = Generator(1)
+generator.load_state_dict(torch.load("./generator.pth", map_location=torch.device('cpu')))
+generator.eval()
+
+
+def generate(seed, num_img):
+    torch.manual_seed(seed)
+    z = torch.randn(num_img, 100, 1, 1)
+    fake_img = generator(z)
+    fake_img = fake_img.detach()
+    fake_img = fake_img.squeeze()
+    save_image(fake_img, "fake_img.png", normalize=True)
+    return 'fake_img.png'
+
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("DCGAN model that generate fake images")
+   
+   
+    image_input = [
+        gr.Slider(0, 1000, label='Seed'),
+        gr.Slider(4, 64, label='Number of images', step=1),
+    ]
+    image_output = gr.Image()
+    image_button = gr.Button("Generate")
+
+    
+    image_button.click(generate, inputs=image_input, outputs=image_output)
+
+demo.launch(share=True)

generator.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:255cba98fd84807b783a17ffdd094f416b524ea0ff9c69904679106914e494d0
+size 14321882

requirement.txt

@@ -0,0 +1,4 @@
+torch
+torchvision
+numpy
+gradio