import torch
from typing import Dict
from PIL import Image
import torch.nn as nn
import pytorch_lightning as pl
from torch_utils import im2tensor
from quickvision.models.detection import detr
from quickvision.models.detection.detr import create_detr_backbone
from quickvision.losses import detr_loss
from quickvision.models.detection.detr import engine
from dataset_utils import DummyDetectionDataset
if(torch.cuda.is_available()):
from torch.cuda import amp
train_dataset = DummyDetectionDataset(img_shape=(3, 256, 256), num_classes=3,
num_samples=10, box_fmt="cxcywh")
val_dataset = DummyDetectionDataset(img_shape=(3, 256, 256), num_classes=3,
num_samples=10, box_fmt="cxcywh")
supported_detr_backbones = ["resnet50", "resnet50_dc5", "resnet101", "resnet101_dc5"]
error_bbone = "invalid_model"
some_supported_backbones = ["resnet50"]
def collate_fn(batch):
return tuple(zip(*batch))
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=2,
shuffle=False, collate_fn=collate_fn)
from PIL import Image
import torch.nn as nn
import pytorch_lightning as pl
from torch_utils import im2tensor
from quickvision.models.detection import detr
from quickvision.models.detection.detr import create_detr_backbone
from quickvision.losses import detr_loss
from quickvision.models.detection.detr import engine
from dataset_utils import DummyDetectionDataset
if (torch.cuda.is_available()):
from torch.cuda import amp
train_dataset = DummyDetectionDataset(img_shape=(3, 256, 256),
num_classes=3,
class_start=0,
normalize=True,
num_samples=10,
box_fmt="cxcywh")
val_dataset = DummyDetectionDataset(img_shape=(3, 256, 256),
num_classes=3,
class_start=0,
normalize=True,
num_samples=10,
box_fmt="cxcywh")
supported_detr_backbones = [
"resnet50", "resnet50_dc5", "resnet101", "resnet101_dc5"
]
error_bbone = "invalid_model"
some_supported_backbones = ["resnet50"]
if (torch.cuda.is_available()):
from torch.cuda import amp
fpn_supported_models = [
"resnet18",
] # "resnet34","resnet50", "resnet101", "resnet152",
# "resnext50_32x4d", "resnext101_32x8d", "wide_resnet50_2", "wide_resnet101_2"
non_fpn_supported_models = ["mobilenet_v2"]
# "resnet18", "resnet34", "resnet50","resnet101",
# "resnet152", "resnext101_32x8d", "mobilenet_v2", "vgg11", "vgg13", "vgg16", "vgg19"
train_dataset = DummyDetectionDataset(img_shape=(3, 256, 256),
num_classes=2,
class_start=1,
num_samples=10,
box_fmt="xyxy")
val_dataset = DummyDetectionDataset(img_shape=(3, 256, 256),
num_classes=2,
class_start=1,
num_samples=10,
box_fmt="xyxy")
def collate_fn(batch):
return tuple(zip(*batch))
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=2,