infer_geco2

GECO2 exemplar-based instance segmentation inference for object counting scenarios.

The algorithm takes one or multiple example bounding boxes (prompts) and predicts object instances as:

• bounding boxes
• confidence scores
• binary masks (one per instance)

🚀 Use with Ikomia API

1. Install Ikomia API

We strongly recommend using a virtual environment. If you're not sure where to start, we offer a tutorial here.

pip install ikomia

2. Create your workflow

from ikomia.dataprocess.workflow import Workflow
from ikomia.utils.displayIO import display

wf = Workflow()
algo = wf.add_task(name="infer_geco2", auto_connect=True)

algo.set_parameters({
    "input_box": "[197.4,21.0,225.3,36.91]"
})

wf.run_on(url="https://raw.githubusercontent.com/jerpelhan/GeCo/refs/heads/main/material/1.jpg")

display(algo.get_image_with_mask_and_graphics())

☀️ Use with Ikomia Studio

Ikomia Studio offers a friendly UI with the same features as the API.

• If you haven't started using Ikomia Studio yet, download and install it from this page.
• For additional guidance on getting started with Ikomia Studio, check out this blog post.

Prompts can be provided in two ways:

• draw one or more rectangles in Studio (graphics input)
• set input_box in parameters

📝 Set algorithm parameters

Parameters:

• model_name: model variant (GECO2FSCD or CNTQG_multitrain_ca44).
• cuda: enable CUDA if available (True/False).
• confidence_threshold: confidence threshold in [0,1].
• input_box: prompt box(es) in JSON:
  • single box: [x1, y1, x2, y2]
  • multiple boxes: [[x1, y1, x2, y2], [x1, y1, x2, y2], ...]

from ikomia.dataprocess.workflow import Workflow
from ikomia.utils.displayIO import display

wf = Workflow()
algo = wf.add_task(name="infer_geco2", auto_connect=True)

algo.set_parameters({
    "model_name": "GECO2FSCD",
    "cuda": "True",
    "confidence_threshold": "0.30",
    "input_box": "[[197.4,21.0,225.3,36.91], [166.1, 147.2, 177.1, 172.1]]"
})

wf.run_on(url="https://raw.githubusercontent.com/jerpelhan/GeCo/refs/heads/main/material/1.jpg")

display(algo.get_image_with_mask_and_graphics())

🔍 Explore algorithm outputs

for output in algo.get_outputs():
    print(output)
    print(output.to_json())