Jetson nano googlenet detect information

andy8902 · May 5, 2020, 7:34am

When I detect the object, the terminal displays object class rate(on the top picture).

The imagenet.py is on bottom picture.Can I use the imagenet.py to change terminal’s inormation ?

AastaLLL · May 5, 2020, 8:56am

Hi,

Not sure which source do you use.
But you can update the output from the source code like this:

dusty-nv/pytorch-classification/blob/master/train.py#L174


      
                                      world_size=args.world_size, rank=args.rank)
          
          
# data loading code
          traindir = os.path.join(args.data, 'train')
          valdir = os.path.join(args.data, 'val')
          normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                           std=[0.229, 0.224, 0.225])
          
          
train_dataset = datasets.ImageFolder(
              traindir,
              transforms.Compose([
                  #transforms.Resize(224),
                  transforms.RandomResizedCrop(args.resolution),
                  transforms.RandomHorizontalFlip(),
                  transforms.ToTensor(),
                  normalize,
              ]))
          
          
num_classes = len(train_dataset.classes)
          print('=> dataset classes:  ' + str(num_classes) + ' ' + str(train_dataset.classes))

Thanks.

andy8902 · May 6, 2020, 3:05am

Tha labels are car,doormax,doormid, and doormin on googlenet.After training googlenet, I used googlenet to detect object. I make GPIO LED(Jetson Nano GPIO - JetsonHacks) with dection.If doormin and doormid recognition rate are closely, I want two leds light togehter.