Creating a variable to track the number of objects detected in a series of images

[matt-humphrey]

I'm developing a model that counts the number of snails on a farm, and have then got a function which will take a static image from google maps and place a marker where each image was taken. I'm trying to link each location/image and the number of snails in that area.

Is it possible to create a variable or a dictionary (or otherwise) which links the number of snails and the image that the count was obtained from, such that I can call that variable after detection?

At the moment, the best method I can think of is using --save-txt when using detect.py, and then parsing through each text file to count the number of snails detected. I'd then create a dictionary with the name of each image as the key, and the number of snails as the value.

Is there a better way to do this?

Cheers

[glenn-jocher]

@matt-humphrey you might want to use YOLOv5 PyTorch Hub models for more customizable python workflows. To count the number of objects in an image:

import torch

# Model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s')

# Images
dir = 'https://github.com/ultralytics/yolov5/raw/master/data/images/'
imgs = [dir + f for f in ('zidane.jpg', 'bus.jpg')]  # batch of images

# Inference
results = model(imgs)
results.print()  # or .show(), .save()

See PyTorch Hub Tutorial to get started:

YOLOv5 Tutorials

• Train Custom Data  🚀 RECOMMENDED
• Weights & Biases Logging  🌟 NEW
• Multi-GPU Training
• PyTorch Hub  ⭐ NEW
• ONNX and TorchScript Export
• Test-Time Augmentation (TTA)
• Model Ensembling
• Model Pruning/Sparsity
• Hyperparameter Evolution
• Transfer Learning with Frozen Layers  ⭐ NEW
• TensorRT Deployment