DeepLabV3.md

Pytorch DeepLabV3+ (Semantic Segmentation)

Environment Setup

Setup AI Model Efficiency Toolkit (AIMET)

Please install and setup AIMET before proceeding further. This model was tested with the torch_gpu variant of AIMET 1.23.

Install dependencies

   python -m pip install pycocotools

Append the repo location to your PYTHONPATH with the following:

export PYTHONPATH=$PYTHONPATH:<path to parent of aimet_model_zoo>

Dataset

The Pascal Dataset can be downloaded from here:

• http://host.robots.ox.ac.uk/pascal/VOC/voc2012/

---

Usage

To run evaluation with QuantSim in AIMET, use the following

python3  aimet_zoo_torch/deeplabv3/evaluators/deeplabv3_quanteval.py \
                --model-config <configuration to be tested> \
                --dataset-path <path to the downloaded Pascal dataset, should end in VOCdevkit/VOC2012> \
                --batch-size  <batch size as an integer value, defaults to 8> \

Available model configurations are:

• dlv3_w4a8
• dlv3_w8a8

---

Model checkpoint and configuration

• The original DeepLabV3+ checkpoint can be downloaded from here:
  • https://drive.google.com/file/d/1G9mWafUAj09P4KvGSRVzIsV_U5OqFLdt/view
• The Quantization Simulation (Quantsim) Configuration file can be downloaded from here: default_config_per_channel.json (Please see this page for more information on this file).

---

Quantization Configuration

W8A8 Quantization | The following configuration has been used:

• Weight quantization: 8 bits, per tensor symmetric quantization
• Bias parameters are not quantized
• Activation quantization: 8 bits, asymmetric quantization
• Model inputs are quantized
• TF-Enhanced was used as quantization scheme
• Cross layer equalization and Adaround has been applied on optimized checkpoint
• Data Free Quantization has been performed on the optimized checkpoint

W4A8 Quantization | The following configuration has been used:

• Weight quantization: 4 bits, per channel symmetric quantization
• Bias parameters are not quantized
• Activation quantization: 8 bits, asymmetric quantization
• Model inputs are quantized
• TF-Enhanced was used as quantization scheme
• Cross layer equalization and Adaround has been applied on optimized checkpoint
• Data Free Quantization has been performed on the optimized checkpoint
• Quantization Aware Traning has been performed on the optimized checkpoint