Please install and setup AIMET before proceeding further.
- Install pycocotools as follows
sudo -H pip install pycocotools
- Clone the DeepLabV3+ repo
git clone https://github.com/jfzhang95/pytorch-deeplab-xception.git
cd pytorch-deeplab-xception
git checkout 9135e104a7a51ea9effa9c6676a2fcffe6a6a2e6
- Apply the following patch to the above repository
git apply ../aimet-model-zoo/zoo_torch/examples/pytorch-deeplab-xception-zoo.patch
- Place modeling directory & dataloaders directory & metrics.py & mypath.py to aimet-model-zoo/zoo_torch/examples/
mv modeling ../aimet-model-zoo/zoo_torch/examples/
mv dataloaders ../aimet-model-zoo/zoo_torch/examples/
mv utils/metrics.py ../aimet-model-zoo/zoo_torch/examples/
mv mypath.py ../aimet-model-zoo/zoo_torch/examples/
- Download Optimized DeepLabV3+ checkpoint from the Releases page.
- Change data location as located in mypath.py
- The original DeepLabV3+ checkpoint can be downloaded here:
- Optimized DeepLabV3+ checkpoint can be downloaded from the Releases page.
- Pascal Dataset can be downloaded here:
- To run evaluation with QuantSim in AIMET, use the following
python eval_deeplabv3.py \
--checkpoint-path <path to optimized checkpoint directory to load from> \
--base-size <base size for Random Crop> \
--crop-size <crop size for Random Crop> \
--num-classes <number of classes in a dataset> \
--dataset <dataset to be used for evaluation> \
--quant-scheme <quantization schme to run> \
--default-output-bw <bitwidth for activation quantization> \
--default-param-bw <bitwidth for weight quantization> - Weight quantization: 8 bits, asymmetric quantization
- Bias parameters are not quantized
- Activation quantization: 8 bits, asymmetric quantization
- Model inputs are not quantized
- TF_enhanced was used as quantization scheme
- Data Free Quantization and Quantization aware Training has been performed on the optimized checkpoint