multi-temporal-crop-classification-baseline
multi-temporal-crop-classification-baseline copied to clipboard
ClarkCGA
Baseline model for crop type segmentation as part of the HLS FM downstream task evaluations
Hi the `default_config` references a dataset: ``` train_dataset_name: chips_filtered_13_classes_complete ``` Can you confirm this is the [exact dataset from HF](https://huggingface.co/datasets/ibm-nasa-geospatial/multi-temporal-crop-classification), and not a modified version (implied by the reference to...
Hi, I have downloaded the model_weights from model_weights/multi_temporal_crop_classification.pth and executed it on the eval_dataset (549 patches) to calculate the accuracy metrics. However, the accuracy measurement is different from the one...
From main readme: However on running `main.ipynb` the metrics are None:
I've had some headaches evaluating on another dataset since the implementation here in `_generate_matrix` assumes nodata is encoded as zero - however in my dataset 0 was a class and...
Setting `gpu_devices=[0, 1, 2, 3]` and calling `compiled_model.fit` I receive: ``` RuntimeError: module must have its parameters and buffers on device cuda:0 (device_ids[0]) but found one of them on device:...
Added this cell, might be useful to include: ```python3 # load previously trained model checkpoint_path = "../output6/Unet_ep100/chkpt/Unet_final_state.pth" checkpoint = torch.load(checkpoint_path, map_location=torch.device('cpu')) # or use GPU # Remove 'module.' prefix if...
From the following config I assume there was a fair bit of experimentation performed to arrive at these parameters - are you able to shed light/into on the experiments run?...