AssertionError problem of Gesturenet inference

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

Hello. When I do the inference of Gesturenet, I got the AssertionError in the process of reading data.json as the picture below.

How should I do to deal with the problem ?

Here is the error message

image
image1920×708 165 KB

Here is the command

sudo docker run -it --rm -v /home/ubuntu/tao_test_2023/gesturenet/:/workspace/tao-experiments/gesturenet/ 
                            nvcr.io/nvidia/tao/tao-toolkit:4.0.1-tf1.15.5 
                            gesturenet inference 
                         -e /workspace/tao-experiments/gesturenet/train_spec.json 
                         -m /workspace/tao-experiments/gesturenet/model.tlt 
                         -k nvidia_tlt 
                         --image_root_path /workspace/tao-experiments/gesturenet/data 
                         --data_json /workspace/tao-experiments/gesturenet/data.json 
                         --data_type kpi_set 
                         --results_dir /workspace/tao-experiments/gesturenet

Here is content of data.json

{    
     "set": "data",    
     "users": {        
        "id14": {},
        "id07": {},
        "id01": {},
        "id04": {},
        "id18": {},
        "id05": {},
        "id11": {},
        "id06": {},
        "id13": {},
        "id08": {},
        "id16": {},
        "id10": {},
        "id09": {},
        "id03": {},
        "id12": {},
        "id02": {},
        "id15": {},
        "id17": {}
    }
}

Here is content of train_spec.json

{
    "random_seed": 108,
    "batch_size": 1,
    "output_experiments_fld": "/workspace/tao-experiments/gesturenet/",
    "save_weights_path": "model",
    "trainer": {
        "class": "ClassifyNetTrainer",
        "module": "driveix.classifynet.trainer.classifynet_trainer",
        "top_training": {
            "stage_order": 1,
            "loss_fn": "categorical_crossentropy",
            "train_epochs": 0, 
            "num_layers_unfreeze": 0,
            "optimizer": "rmsprop"
        },
        "finetuning": {
            "stage_order": 2,
            "train_epochs": 50,
            "loss_fn": "categorical_crossentropy",
            "initial_lrate": 5e-05,
            "decay_step_size": 33,
            "lr_drop_rate": 0.5,
            "enable_checkpointing": true,
            "num_layers_unfreeze": 3,
            "optimizer": "sgd"
        }, 
        "num_workers": 1
    },
    "model": {
        "image_height": 160,
        "image_width": 160,
        "gray_scale_input": false,
        "data_format": "channels_first",
        "base_model": "resnet_vanilla",
        "num_layers": 18,
        "use_batch_norm": true,
        "weights_init": "/workspace/tao-experiments/gesturenet/model.tlt",
        "add_new_head": false,
        "kernel_regularizer_type": "l2",
        "kernel_regularization_factor": 0.001
    },
    "dataset": {
        "image_root_path": "/workspace/tao-experiments/gesturenet/data",
        "classes": {
            "thumbs_up": 0,
            "fist": 1,
            "stop": 2,
            "ok": 3,
            "two": 4,
            "random": 5
        },
        "data_path": "/workspace/tao-experiments/gesturenet/data.json",
        "num_classes": 6,
        "augmentation": {
            "shear_range": 0.0,
            "color_pca_aug": {
                "enable": false,
                "probability": 0.5
            },
            "gamma_aug": {
                "enable": true,
                "probability": 0.5,
                "lower_limit": 0.5,
                "upper_limit": 2.0
            },
            "rotation_range": 5,
            "brightness_range": [
                0.5,
                1.5
            ], 
            "occlusion_aug": {
                "max_aspect_ratio": 3.33,
                "max_area": 0.25,
                "enable": true,
                "probability": 0.5,
                "pixel_level": true,
                "min_area": 0.05,
                "min_pixel": 0,
                "max_pixel": 255,
                "min_aspect_ratio": 0.3
            },
            "horizontal_flip": true
        }
    },
    "evaluator": {
        "evaluation_exp_name": "results", 
        "data_path": "/workspace/tao-experiments/gesturenet/data.json"
    }
}

Thank you for your help in advance.

2

The data_type should be in data json file.

4

Excuse me. @Morganh Do you mean that the format of data.json should be like this ?

{    
     "set": "data",
     "data_type": "kpi_set", 
     "users": {
5

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Yes, it is.
The assertion error comes from “assert data_type in data_json”

6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.