Optimization API¶
optimizer/convert_compare.py: To convert a pytorch model to a tflite model and compare the difference between their outputs
Match a NAS model to a original model
fix_nasyolo, fix_nasbackbone
Match two pytorch models with same structure, but different hyperparameters (ex: input channels, output channels)
match_nas, match
Extract Yolov5’s backbone from an original model
extract_backbone
optimizer/model_evaluation.py: Evaluate the inference time, network parameters and flops of a specific model
optimizer/loss.py: Calculating the loss between teacher model and student model
optimizer/prune.py: Call the function prune to prune models