Pytorch train a model for object detection in an image

This model detect only 2 object bottle and car

Step 1: Description

• At first you need to decided how may object you need to detect. Here i used 2 object car and bottle
• Object name write into a json 'cat_to_name.json' file {"1":"bottle","2":"Car"}
• We need to arrange/collect train image. all image will be in data folder like this. data/train/1/images_00000.jpg here 1 for all bottle image and another data/train/2/images_00000.jpg car images
• if you use google colab then open object_detect_pytorch.ipynb
• zip you data directory
• upload to colad file system
• unzip data folder
• upload json file cat_to_name.json
• then run object_detect_pytorch.ipynb ste

Step 2: Initialize

#cat_to_name_file
{"1":"bottle","2":"Car"}
#Change some parameter it's depend on imput class (we used 2) and output (we used 2) 
num_features = model.classifier.in_features
classifier = nn.Sequential(OrderedDict([
                              ('fc1', nn.Linear(num_features, 512)), #### change input
                              ('relu', nn.ReLU()),
                              ('drpot', nn.Dropout(p=0.5)),
                              ('hidden', nn.Linear(512, 100)),                       
                              ('fc2', nn.Linear(100, 2)), ##### Change output
                              ('output', nn.LogSoftmax(dim=1)),
                              ]))

model.classifier = classifier    

Save the checkpoint

# TODO: Save the checkpoint 
model.class_to_idx = dataloaders['train'].dataset.class_to_idx
model.epochs = num_epochs
checkpoint = {'input_size': [3, 224, 224],
                 'batch_size': dataloaders['train'].batch_size,
                  'output_size': 2,
                  'state_dict': model.state_dict(),
                  'optimizer_dict':optimizer.state_dict(),
                  'class_to_idx': model.class_to_idx,
                  'epoch': model.epochs}
torch.save(checkpoint, 'object-detect.pth')    

So we get a pth file object-detect.pth file now we are ready to test our model

python predict.py    

None issue

There are no object-detect.pth file. so we need to train and save the file.