Mobile Object Localizer

Our new model ZOO works with DepthAI V3. Find out more in our documentation.

Model Details

Model Description

MobileObjectLocalizer is a general-purpose object detection model developed by Google that can be used for any type of object. Unlike models such as YOLO which classifies objects among the predefined set of classes, MobileObjectLocalizer can detect any object. and does not assign any category.

Developed by: Google

Shared by:

Model type: Computer vision

License:

Training Details

No training details available.

Testing Details

No testing details available.

Technical Specifications

Input/Output Details

Input:

Name: normalized_input_image_tensor

Info: NCHW BGR un-normalized image

Output:

Name: scores and bboxes

Info: Scores and bounding boxes that still need NMS.

Model Architecture

Backbone: MobileNetV2 backbone with a 0.75 width-multiplier,

Head: SSDLite detection head.

Throughput

Model variant: mobile-object-localizer:192x192

• Input shape: [1, 3, 192, 192] • Output shapes: [[1, 1, 738], [1, 738, 4]]

• Params (M): 1.746 • GFLOPs: 0.175

Platform	Precision	Throughput (infs/sec)	Power Consumption (W)
RVC2	FP16	67.74	N/A
RVC4	FP16	603.80	2.85

* Benchmarked with , using 2 threads (and the DSP runtime in balanced mode for RVC4).

* Parameters and FLOPs are obtained from the package.

Utilization

Models converted for RVC Platforms can be used for inference on OAK devices. DepthAI pipelines are used to define the information flow linking the device, inference model, and the output parser (as defined in model head(s)). Below, we present the most crucial utilization steps for the particular model. Please consult the docs for more information.

Install DepthAI and depthai-nodes:

pip install --upgrade depthai depthai-nodes

selects native DepthAI parsers from the NN Archive. Use for the Python parser implementations; both return .

Define model:

import depthai as dai
from depthai_nodes.node import ParsingNeuralNetwork

model_description = dai.NNModelDescription(
    "luxonis/mobile-object-localizer:192x192"
)

nn = pipeline.create(ParsingNeuralNetwork).build(
    <CameraNode>, model_description
)

Model heads (parser names in the NN Archive):

head (uses dai.node.DetectionParser by default) that outputs message with bounding boxes and scores.

Get parsed output(s):

parser_output_queue = nn.out.createOutputQueue()
pipeline.start()

while pipeline.isRunning():
    parser_output: dai.ImgDetections = parser_output_queue.get()

Example

You can quickly run the model using our script. It automatically downloads the model, creates a DepthAI pipeline, runs the inference, and displays the results using our DepthAI visualizer tool. To try it out, run:

python3 main.py \
    --model luxonis/mobile-object-localizer:192x192

A class-agnostic mobile object detector.
License	Apache 2.0 Commercial use
Downloads	2011
Tasks	Object Detection
Model Types	ONNX

Model Variants

Name	Version	Available For	Created At	Deploy
		RVC2, RVC4	4 days ago