automl
automl copied to clipboard
google
Google Brain AutoML
Addressing issue in https://github.com/google/automl/issues/1171
I'm running the TfLite model analyzer on the EfficientDet TfLite file like so: ``` model_path = "/content/efficientdet.tflite" tf.lite.experimental.Analyzer.analyze(model_path=model_path, gpu_compatibility=True) ``` One [output line](https://gist.github.com/saad-palapa/a3827a9ed6271db217a4b59c2498e602#file-gistfile1-txt-L491) shows that the last layer is not...
I have converted the SKU110K dataset in to tfrecord format and when I train it on efficeintdet model, it results in below error. I have attached my tfrecord conversion script...
python3.9 Successfully installed keras-efficientnet-v2-1.2.2 tensorflow-2.9.1 when I import effnetv2_model the result is: ModuleNotFoundError: No module named 'effnetv2_model' how to solve the problem?
Hi, there is the problem : penta@dell-r740:~$ python3 automl/efficientdet/tf2/train.py --train_file_pattern=train_tf.tfrecord --val_file_pattern=val_tf.tfrecord --model_name=efficientdet-d0 --model_dir=/tmp/efficientdet-d0 -finetune --pretrained_ckpt=efficientdet-d0 --batch_size=24 --eval_samples=1024 --num_examples_per_epoch=33707 --num_epochs=50 --hparams=voc_config.yaml --strategy=gpus 2024-02-13 15:12:30.743921: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on....
I would like to fine-tune a pretrained efficientdet-lite0 model on my dataset and then do post training quantization (full int8). What is the recommended way of doing this? I did...
Why in one case you use the outputs for class and box predictions in other case you use detection for the same ?
I am currently using efficient det lite family to train on a custom dataset but the data is imbalanced. I alsready worked with an "other" class but the results are...
This function already does summation in place, and therefore errors out when the inplace parameters is passed into the _add func.
