DLA (Deep Layer Aggregation)

Use case : Image classification

Model description

Deep Layer Aggregation (DLA) networks introduce iterative and hierarchical structures for aggregating features across layers. DLA extends standard architectures by merging features from different depths and resolutions, enabling richer semantic and spatial information flow.

DLA employs Hierarchical Deep Aggregation (HDA) to merge feature hierarchies combining features from different depths, and Iterative Deep Aggregation (IDA) to progressively refine resolution and semantic information. The dense connections enable gradient flow and feature reuse across the network.

DLA is particularly well-suited for applications requiring multi-scale feature representation, such as semantic segmentation and object detection.

(source: https://arxiv.org/abs/1707.06484)

The model is quantized to int8 using ONNX Runtime and exported for efficient deployment.

Network information

Network inputs / outputs

For an image resolution of NxM and P classes

Recommended platforms

Performances

Metrics

• Measures are done with default STEdgeAI Core configuration with enabled input / output allocated option.
• All the models are trained from scratch on Imagenet dataset

Reference NPU memory footprint on Imagenet dataset (see Accuracy for details on dataset)

Reference NPU inference time on Imagenet dataset (see Accuracy for details on dataset)

Accuracy with Imagenet dataset

Dataset details: link Number of classes: 1000. To perform the quantization, we calibrated the activations with a random subset of the training set. For the sake of simplicity, the accuracy reported here was estimated on the 50000 labelled images of the validation set.

Retraining and Integration in a simple example:

Please refer to the stm32ai-modelzoo-services GitHub here

References

[1] - Dataset: Imagenet (ILSVRC 2012) — https://www.image-net.org/

[2] - Model: Deep Layer Aggregation — https://github.com/ucbdrive/dla