# Get all files from the dataset.
print("Find all files for evaluation...")
pickle_files = glob.glob(os.path.join(dataset_path, "**", "*.p"))
# Evaluate all files.
# TODO parallelize this.
print("Evaluate all files...")
data = {"results": []}
for index, pickle_file in enumerate(pickle_files):
name = os.path.basename(pickle_file).split(".")[0]
# Load and preprocess the data.
pointcloud, targets = pickle.load(open(pickle_file, "rb"))
pointcloud = np.array(
[preprocess_pointcloud(pointcloud, 1024, list(range(3)))])
#targets = preprocess_targets(targets, [0])[0] # 0 is height
targets = preprocess_targets(targets, [1])[0] # 1 is weight
predicted_targets = model.predict(pointcloud)[0][0]
# Store results.
result = {
"name": name,
"targets": str(targets),
"predicted_targets": str(predicted_targets),
"error": str(np.abs(predicted_targets - targets))
}
data["results"].append(result)
if index % 1000 == 0:
print("{} per cent".format(int(100 * (index / len(pickle_files)))))
def py_load_pickle(path):
pointcloud, targets = pickle.load(open(path.numpy(), "rb"))
pointcloud = preprocess_pointcloud(pointcloud, subsample_size,
channels)
targets = preprocess_targets(targets, targets_indices)
return pointcloud, targets