SSD === .. _ssd: Preparing the Dataset --------------------------- The dataset for SSD contains images and corresponding label files in KITTI text label format. The data structure must be in the following format: .. code:: /Dataset_01 /images 0000.jpg 0001.jpg 0002.jpg ... ... ... N.jpg /labels 0000.txt 0001.txt 0002.txt ... ... ... N.txt Creating a Configuration File ----------------------------- .. _creating_a_configuration_file_ssd: Below is a sample of the SSD spec file. It has six major components: :code:`ssd_config`, :code:`training_config`, :code:`eval_config`, :code:`nms_config`, :code:`augmentation_config`, and :code:`dataset_config`. The format of the spec file is a protobuf text(prototxt) message and each of its fields can be either a basic data type or a nested message. .. code:: random_seed: 42 ssd_config { aspect_ratios: "[[1.0, 2.0, 0.5], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5, 3.0, 1.0/3.0], [1.0, 2.0, 0.5], [1.0, 2.0, 0.5]]" scales: "[0.07, 0.15, 0.33, 0.51, 0.69, 0.87, 1.05]" two_boxes_for_ar1: true clip_boxes: false variances: "[0.1, 0.1, 0.2, 0.2]" arch: "resnet" nlayers: 18 freeze_bn: false freeze_blocks: 0 } training_config { batch_size_per_gpu: 16 num_epochs: 80 enable_qat: false learning_rate { soft_start_annealing_schedule { min_learning_rate: 5e-5 max_learning_rate: 2e-2 soft_start: 0.15 annealing: 0.8 } } regularizer { type: L1 weight: 3e-5 } } eval_config { validation_period_during_training: 10 average_precision_mode: SAMPLE batch_size: 16 matching_iou_threshold: 0.5 } nms_config { confidence_threshold: 0.01 clustering_iou_threshold: 0.6 top_k: 200 } augmentation_config { output_width: 300 output_height: 300 output_channel: 3 } dataset_config { data_sources: { label_directory_path: "/path/to/train/labels" image_directory_path: "/path/to/train/images" } include_difficult_in_training: true target_class_mapping { key: "car" value: "car" } target_class_mapping { key: "pedestrian" value: "pedestrian" } target_class_mapping { key: "cyclist" value: "cyclist" } target_class_mapping { key: "van" value: "car" } target_class_mapping { key: "person_sitting" value: "pedestrian" } validation_data_sources: { label_directory_path: "/path/to/val/labels" image_directory_path: "/path/to/val/images" } } The top level structure of the spec file is summarized in the sections below. Training Config ^^^^^^^^^^^^^^^ The training configuration (:code:`training_config`) defines the parameters needed for the training, evaluation, and inference. Details are summarized in the table below. +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | batch_size_per_gpu | The batch size for each GPU, so the effective batch size is batch_size_per_gpu * num_gpus | Unsigned int, positive | -- | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | num_epochs | The anchor batch size used to train the RPN | Unsigned int, positive | -- | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | enable_qat | Whether to use quantization-aware training | Boolean | -- | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | learning_rate | Only soft_start_annealing_schedule with these nested parameters is supported. | Message type | -- | | | | | | | | 1. min_learning_rate: The minimum learning during the entire experiment | | | | | 2. max_learning_rate: The maximum learning during the entire experiment | | | | | 3. soft_start: Time to lapse before warm up ( expressed in percentage of progress | | | | | between 0 and 1) | | | | | 4. annealing: Time to start annealing the learning rate | | | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ | regularizer | This parameter configures the regularizer to be used while training and contains the | Message type | L1 | | | following nested parameters. | | | | | | | **Note**: NVIDIA suggests using the L1 regularizer when training a network | | | 1. type: The type or regularizer to use. NVIDIA supports NO_REG, L1, and L2 | | before pruning as L1 regularization helps make the network weights more | | | 2. weight: The floating point value for the regularizer weight | | prunable. | +--------------------+-------------------------------------------------------------------------------------------+-------------------------------+-------------------------------------------------------------------------------+ .. Note:: The learning rate is automatically scaled with the number of GPUs used during training, or the effective learning rate is :code:`learning_rate * n_gpu`. Evaluation Config ^^^^^^^^^^^^^^^^^ The evaluation configuration (:code:`eval_config`) defines the parameters needed for the evaluation either during training or as a standalone procedure. Details are summarized in the table below. +-----------------------------------+----------------------------------------------------------------------------------------------+----------------------------------+-------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +-----------------------------------+----------------------------------------------------------------------------------------------+----------------------------------+-------------------------------+ | validation_period_during_training | The number of training epochs per validation | Unsigned int, positive | 10 | +-----------------------------------+----------------------------------------------------------------------------------------------+----------------------------------+-------------------------------+ | average_precision_mode | The Average Precision (AP) calculation mode can be either SAMPLE or INTEGRATE. SAMPLE | ENUM type ( SAMPLE or INTEGRATE) | SAMPLE | | | is used as VOC metrics for VOC 2009 or before. INTEGRATE is used for VOC 2010 or after. | | | +-----------------------------------+----------------------------------------------------------------------------------------------+----------------------------------+-------------------------------+ | matching_iou_threshold | The lowest IoU of the predicted box and ground truth box that can be considered a match. | Boolean | 0.5 | +-----------------------------------+----------------------------------------------------------------------------------------------+----------------------------------+-------------------------------+ NMS Config ^^^^^^^^^^ The NMS configuration (:code:`nms_config`) defines the parameters needed for NMS postprocessing. The NMS configuration applies to the NMS layer of the model in training, validation, evaluation, inference, and export. Details are summarized in the table below. +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | confidence_threshold | Boxes with a confidence score less than confidence_threshold are discarded before applying NMS | float | 0.01 | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | cluster_iou_threshold | IoU threshold below which boxes will go through the NMS process | float | 0.6 | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | top_k | top_k boxes will be outputted after the NMS keras layer. If the number of valid boxes is less | Unsigned int | 200 | | | than k, the returned array will be padded with boxes whose confidence score is 0. | | | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ Augmentation Config ^^^^^^^^^^^^^^^^^^^ The :code:`augmentation_config` parameter defines the image size after preprocessing. The augmentation methods in the SSD `paper`_ will be performed during training, including random flip, zoom-in, zoom-out and color jittering. And the augmented images will be resized to the output shape defined in :code:`augmentation_config`. In evaluation process, only the resize will be performed. .. Note:: The details of augmentation methods can be found in setcion 2.2 and 3.6 of the `paper`_. .. _paper: https://arxiv.org/abs/1512.02325v5 +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | output_width | The width of preprocessed images and the network input | Unsigned int | 300 | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | output_height | The height of preprocessed images and the network input | Unsigned int | 300 | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | output_channel | The channel of preprocessed images | Unsigned int | 3 | +-----------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ Dataset Config ^^^^^^^^^^^^^^ The :code:`dataset_config` parameter defines the path to the training dataset, validation dataset, and target_class_mapping. +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | data_sources | The path to the training dataset images and labels. | Message type | | | | | | | | | | | | | | | | | | | | * label_directory_path: The path to the label directory | | | | | * image_directory_path: The path to the image directory | | | +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | include_difficult_in_training | Specifies whether to include difficult objects in the label (the Pascal VOC difficult label | bool | true | | | or KITTI occluded objects) | | | +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | validation_data_sources | The path to the training dataset images and labels | Message type | | +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ | target_class_mapping | A mapping of classes in labels to the target classes | Message type | | +---------------------------------+------------------------------------------------------------------------------------------------+-------------------------------+-------------------------------+ .. Note:: :code:`data_sources` and :code:`validation_data_sources` are both repeated fields. Multiple datasets can be added to sources. SSD config ^^^^^^^^^^ The SSD configuration (:code:`ssd_config`) defines the parameters needed for building the SSD model. Details are summarized in the table below. +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | **Field** | **Description** | **Data Type and Constraints** | **Recommended/Typical Value** | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | aspect_ratios_global | The anchor boxes of aspect ratios defined in aspect_ratios_global will be generated | string | “[1.0, 2.0, 0.5, 3.0, 0.33]” | | | for each feature layer used for prediction. Note that either the aspect_ratios_global or | | | | | aspect_ratios parameter is required; you don't need to specify both. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | aspect_ratios | The aspect ratio of anchor boxes for different SSD feature layers | string | “[[1.0,2.0,0.5], | | | | | [1.0,2.0,0.5], | | | | | [1.0,2.0,0.5], | | | **Note**: Either the aspect_ratios_global or aspect_ratios parameter is required; you don't need to specify both. | | [1.0,2.0,0.5], | | | | | [1.0,2.0,0.5], | | | | | [1.0, 2.0, 0.5, 3.0, 0.33]]” | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | two_boxes_for_ar1 | If this parameter is True, two boxes will be generated with an aspect ratio of 1. | Boolean | True | | | | | | | | One with a scale for this layer and the other with a scale that is the geometric mean of the scale | | | | | | | | | | for this layer and the scale for the next layer. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | clip_boxes | If true, all corner anchor boxes will be truncated so they are fully inside the feature images. | Boolean | False | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | scales | A list of positive floats containing scaling factors per convolutional predictor layer. | string | “[0.05, 0.1, 0.25, 0.4, 0.55, 0.7, 0.85]” | | | This list must be one element longer than the number of predictor layers so that, if two_boxes_for_ar1 is | | | | | true, the second aspect-ratio 1.0 box for the last layer can have a proper scale. Except for the last | | | | | element in this list, each positive float is the scaling factor for boxes in that layer. For example, | | | | | if for one layer the scale is 0.1, then the generated anchor box with aspect ratio 1 for that layer | | | | | (the first aspect-ratio 1 box if two_boxes_for_ar1 is set to True) will have its height and width as | | | | | 0.1*min(img_h, img_w). | | | | | | | | | | min_scale and max_scale are two positive floats. If both of them appear in the config, the program | | | | | can automatically generate the scales by evenly splitting the space between min_scale and max_scale. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | min_scale/max_scale | If both appear in the config, scales will be generated evenly by splitting the space between min_scale and | float | - | | | max_scale. | | | | variances | A list of 4 positive floats. The four floats, in order, represent variances for box | | | | | center x, box center y, log box height, and log box width. The box offset for box center (cx, cy) and log box | | | | | size (height/width) w.r.t. anchor will be divided by their respective variance value. Therefore, larger | | | | | variances result in less significant differences between two different boxes on encoded offsets. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | steps | An optional list inside quotation marks with a length that is the number of feature layers for prediction. The | string | - | | | elements should be floats or tuples/lists of two floats. The steps define how many pixels apart the anchor-box | | | | | center points should be. If the element is a float, both vertical and horizontal margin is the same. Otherwise, | | | | | the first value is step_vertical and the second value is step_horizontal. If steps are not provided, anchor | | | | | boxes will be distributed uniformly inside the image. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | offsets | An optional list of floats inside quotation marks with length equal to the number of feature layers for prediction. | string | - | | | The first anchor box will have a margin of offsets[i]*steps[i] pixels from the left and top borders. If offsets are | | | | | not provided, 0.5 will be used as default value. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | arch | The backbone for feature extraction. Currently, “resnet”, “vgg”, “darknet”, “googlenet”, “mobilenet_v1”, | string | resnet | | | “mobilenet_v2” and “squeezenet” are supported. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | nlayers | The number of conv layers in a specific arch. For “resnet”, 10, 18, 34, 50 and 101 are supported. For “vgg”, 16 and | Unsigned int | - | | | 19 are supported. For “darknet”, 19 and 53 are supported. All other networks don’t have this configuration, and | | | | | users should delete this parameter from the config file. | | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | freeze_bn | Whether to freeze all batch normalization layers during training. | boolean | False | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ | freeze_blocks | The list of block IDs to be frozen in the model during training. You can choose to freeze some of the CNN blocks | list(repeated integers) | - | | | in the model to make the training more stable and/or easier to converge. The definition of a block is heuristic | | | | | for a specific architecture. For example, by stride or by logical blocks in the model, etc. However, the block ID | • ResNet series. For the ResNet series, the block IDs valid | | | | numbers identify the blocks in the model in a sequential order so you don't have to know the exact locations of the | for freezing is any subset of [0, 1, 2, 3] (inclusive) | | | | blocks when you do training. As a general principle, the smaller the block ID, the closer it is to | • VGG series. For the VGG series, the block IDs valid for | | | | the model input; the larger the block ID, the closer it is to the model output. | freezing is any subset of[1, 2, 3, 4, 5] (inclusive) | | | | | • GoogLeNet. For the GoogLeNet, the block IDs valid for freezing | | | | You can divide the whole model into several blocks and optionally freeze a subset of it. Note that for FasterRCNN, | is any subset of[0, 1, 2, 3, 4, 5, 6, 7] (inclusive) | | | | you can only freeze the blocks that are before the ROI pooling layer. Any layer after the ROI pooling layer will | • MobileNet V1. For the MobileNet V1, the block IDs valid for freezing | | | | not be frozen anyway. For different backbones, the number of blocks and the block ID for each block are different. | is any subset of [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11] (inclusive) | | | | It deserves some detailed explanations on how to specify the block IDs for each backbone. | • MobileNet V2. For the MobileNet V2, the block IDs valid for freezing | | | | | is any subset of [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13] (inclusive) | | | | | • DarkNet. For the DarkNet 19 and DarkNet 53, the block IDs valid for freezing | | | | | is any subset of [0, 1, 2, 3, 4, 5] (inclusive) | | +----------------------+-----------------------------------------------------------------------------------------------------------------------------+--------------------------------------------------------------------------------+-------------------------------------------+ Training the Model ------------------ Train the SSD model using this command: .. code:: tlt ssd train -e -r -k [--gpus ] [--gpu_index ] [--use_amp] [--log_file ] [-m ] [--initial_epoch ] Required Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-r, --results_dir:code:`: Path to the folder where the experiment output is written. * :code:`-k, --key`: Provide the encryption key to decrypt the model. * :code:`-e, --experiment_spec_file`: Experiment specification file to set up the evaluation experiment. This should be the same as the training specification file. Optional Arguments ^^^^^^^^^^^^^^^^^^ * :code:`--gpus num_gpus`: Number of GPUs to use and processes to launch for training. The default = 1. * :code:`--gpu_index`: The GPU indices used to run the training. We can specify the GPU indices used to run training when the machine has multiple GPUs installed. * :code:`--use_amp`: A flag to enable AMP training. * :code:`--log_file`: The path to the log file. Defaults to :code:`stdout`. * :code:`-m, --resume_model_weights`: Path to a pre-trained model or model to continue training. * :code:`--initial_epoch`: Epoch number to resume from. * :code:`--use_multiprocessing`: Enable multiprocessing mode in data generator. * :code:`-h, --help`: Show this help message and exit. Sample Usage ^^^^^^^^^^^^ Here's an example of using the train command on an SSD model: .. code:: tlt ssd train --gpus 2 -e /path/to/spec.txt -r /path/to/result -k $KEY Evaluating the Model -------------------- Use the following command to run evaluation for an SSD model: .. code:: tlt ssd evaluate -m -e [-k ] [--gpu_index ] [--log_file ] Required Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-m, --model`: The :code:`.tlt` model or :code:`TRT` engine to be evaluated. * :code:`-e, --experiment_spec_file`: The experiment spec file to set up the evaluation experiment. This should be the same as the training spec file. Optional Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-h, --help`: Show this help message and exit. * :code:`-k, --key`：The encoding key for the :code:`.tlt` model * :code:`--gpu_index`: The index of the GPU to run evaluation (useful when the machine has multiple GPUs installed). Note that evaluation can only run on a single GPU. * :code:`--log_file`: The path to the log file. The default path is :code:`stdout`. Here is a sample command to evaluate a SSD model: .. code:: tlt ssd evaluate -m /path/to/trained_tlt_ssd_model -k -e /path/to/ssd_spec.txt Running Inference on the Model ------------------------------ The :code:`inference` command for SSD networks can be used to visualize bboxes or generate frame-by-frame KITTI format labels on a directory of images. Here's an example of using this tool: .. code:: tlt ssd inference -i -o -e -m -k [-l ] [-t ] [--gpu_index ] [--log_file ] Required Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-m, --model`: The path to the pretrained model (TLT model). * :code:`-i, --in_image_dir`: The directory of input images for inference. * :code:`-o, --out_image_dir`: The directory path to output annotated images. * :code:`-k, --key`: The key to the load model. * :code:`-e, --config_path`: The path to an experiment spec file for training. Optional Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-t, --draw_conf_thres`: The threshold for drawing a bbox (default: 0.3). * :code:`-h, --help`: Show this help message and exit. * :code:`-l, --out_label_dir`: The directory to output KITTI labels to. * :code:`--gpu_index`: The index of the GPU to run evaluation (useful when the machine has multiple GPUs installed). Note that evaluation can only run on a single GPU. * :code:`--log_file`: The path to the log file. The default path is :code:`stdout`. Here is a sample of using inference with the SSD model: .. code:: tlt ssd inference -i /path/to/input/images_dir -o /path/to/output/dir -m /path/to/trained_tlt_ssd_model -k -e /path/to/ssd_spec.txt Pruning the Model ----------------- .. _pruning_the_model_ssd: Pruning removes parameters from the model to reduce the model size without compromising the integrity of the model itself. The :code:`prune` command includes these parameters: .. code:: tlt ssd prune [-h] -m -o -k [-n ] [-eq ] [-pg ] [-pth ] [-nf ] [-el [] [--gpu_index ] [--log_file ] Required Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-m, --pretrained_model`: The path to the pretrained model. * :code:`-o, --output_file`: The path to output checkpoints to. * :code:`-k, --key`: The key to load a :code:`.tlt` model. Optional Arguments ^^^^^^^^^^^^^^^^^^ * :code:`-h, --help`: Show this help message and exit. * :code:`-n, –normalizer`: ``max`` to normalize by dividing each norm by the maximum norm within a layer; ``L2`` to normalize by dividing by the L2 norm of the vector comprising all kernel norms. (default: `max`) * :code:`-eq, --equalization_criterion`: Criteria to equalize the stats of inputs to an element wise op layer, or depth-wise convolutional layer. This parameter is useful for resnets and mobilenets. Options are :code:`arithmetic_mean`, :code:`geometric_mean`, :code:`union`, and :code:`intersection`. (default: :code:`union`) * :code:`-pg, -pruning_granularity`: Number of filters to remove at a time. (default:8) * :code:`-pth`: Threshold to compare normalized norm against. (default:0.1) .. Note: NVIDIA recommends changing the threshold to keep the number of parameters in the model to within 10-20% of the original unpruned model. * :code:`-nf, --min_num_filters`: Minimum number of filters to keep per layer (default:16) * :code:`-el, --excluded_layers`: List of excluded_layers. Examples: -i item1 item2 (default: []) * :code:`--gpu_index`: The index of the GPU to run evaluation (useful when the machine has multiple GPUs installed). Note that evaluation can only run on a single GPU. * :code:`--log_file`: The path to the log file. Defaults to :code:`stdout`. Here's an example of using the :code:`prune` command: .. code:: tlt ssd prune -m /workspace/output/weights/resnet_003.tlt \ -o /workspace/output/weights/resnet_003_pruned.tlt \ -eq union \ -pth 0.7 -k $KEY After pruning, the model needs to be retrained. See :ref:`Re-training the Pruned Model ` for more details. Re-training the Pruned Model ---------------------------- .. _re-training_the_pruned_model_ssd: Once the model has been pruned, there might be a slight decrease in accuracy. This happens because some previously useful weights may have been removed. To regain accuracy, NVIDIA recommends that you retrain this pruned model over the same dataset. To do this, use the :code:`tlt ssd train` command with an updated spec file that points to the newly pruned model as the pretrained model file. Users are advised to turn off the regularizer in the :code:`training_config` for detectnet to recover the accuracy when retraining a pruned model. You may do this by setting the regularizer type to NO_REG, as mentioned :ref:`here`. All the other parameters may be retained in the spec file from the previous training. Exporting the Model ------------------- .. _exporting_the_model_ssd: The Transfer Learning Toolkit includes the :code:`export` command to export and prepare TLT models for :ref:`Deploying to DeepStream `. The :code:`export` command optionally generates the calibration cache for TensorRT INT8 engine calibration. Exporting the model decouples the training process from inference and allows conversion to TensorRT engines outside the TLT environment. TensorRT engines are specific to each hardware configuration and should be generated for each unique inference environment. This may be interchangeably referred to as the :code:`.trt` or :code:`.engine` file. The same exported TLT model may be used universally across training and deployment hardware. This is referred to as the :code:`.etlt` file or encrypted TLT file. During model export, the TLT model is encrypted with a private key. This key is required when you deploy this model for inference. INT8 Mode Overview ^^^^^^^^^^^^^^^^^^ TensorRT engines can be generated in INT8 mode to improve performance, but require a calibration cache at engine creation-time. The calibration cache is generated using a calibration tensor file, if :code:`export` is run with the :code:`--data_type` flag set to :code:`int8`. Pre-generating the calibration information and caching it removes the need for calibrating the model on the inference machine. Moving the calibration cache is usually much more convenient than moving the calibration tensorfile, since it is a much smaller file and can be moved with the exported model. Using the calibration cache also speeds up engine creation as building the cache can take several minutes to generate depending on the size of the Tensorfile and the model itself. The export tool can generate an INT8 calibration cache by ingesting training data using the following method: * Pointing the tool to a directory of images that you want to use to calibrate the model. For this option, make sure to create a sub-sampled directory of random images that best represent your training dataset. FP16/FP32 Model ^^^^^^^^^^^^^^^ The :code:`calibration.bin` is only required if you need to run inference at INT8 precision. For FP16/FP32-based inference, the export step is much simpler: all you need to do is provide a model from the :code:`train` step to :code:`export` to convert it into an encrypted TLT model. .. image:: ../../content/fp16_fp32_export.png Exporting command ^^^^^^^^^^^^^^^^^ Use the following command to export an SSD model .. code:: tlt ssd export [-h] -m -k [-o ] [--cal_data_file ] [--cal_image_dir ] [--data_type ] [--batches ] [--max_batch_size ] [--max_workspace_size ] [--experiment_spec ] [--engine_file ] [--verbose Verbosity of the logger] [--force_ptq Flag to force PTQ] Required Arguments ****************** * :code:`-m, --model`: The path to the :code:`.tlt` model file to be exported using :code:`export`. * :code:`-k, --key`: The key used to save the :code:`.tlt` model file. * :code:`-e, --experiment_spec`: The path to the spec file. Optional Arguments ****************** * :code:`-o, --output_file`: The path to save the exported model to. The default path is :code:`./.etlt`. * :code:`--data_type`: The desired engine data type. The options are "fp32", "fp16", "int8". The default value is "fp32". If using int8, the following INT8 arguments are required. * :code:`-s, --strict_type_constraints`: A Boolean flag to indicate whether or not to apply the TensorRT :code:`strict_type_constraints` when building the TensorRT engine. Note this is only for applying the strict type of INT8 mode. INT8 Export Mode Required Arguments ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ * :code:`--cal_data_file`: The tensorfile generated from :code:`tlt-int8-tensorfile` for calibrating the engine. This can also be an output file if used with :code:`--cal_image_dir`. * :code:`--cal_image_dir`: The directory of images to use for calibration. .. Note:: The :code:`--cal_image_dir` parameter applies the necessary preprocessing to generate a tensorfile at the path mentioned in the :code:`--cal_data_file` parameter, which is in turn used for calibration. The number of generated batches in the tensorfile is obtained from the value set to the :code:`--batches` parameter, and the :code:`batch_size` is obtained from the value set to the :code:`--batch_size` parameter. Ensure that the directory mentioned in :code:`--cal_image_dir` has at least :code:`batch_size * batches` number of images in it. The valid image extensions are :code:`.jpg`, :code:`.jpeg`, and :code:`.png`. In this case, the :code:`input_dimensions` of the calibration tensors are derived from the input layer of the :code:`.tlt` model. INT8 Export Optional Arguments ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ * :code:`--cal_cache_file`: The path to save the calibration cache file to. The default value is :code:`./cal.bin`. * :code:`--batches`: The number of batches to use for calibration and inference testing. The default value is 10. * :code:`--batch_size`: The batch size to use for calibration. The default value is 8. * :code:`--max_batch_size`: The maximum batch size of the TensorRT engine. The default value is 16. * :code:`--max_workspace_size`: The maximum workspace size of the TensorRT engine. The default value is 1073741824 = 1<<30. * :code:`--experiment_spec`: The experiment_spec for training/inference/evaluation. This is used to generate the graphsurgeon config script for FasterRCNN from the experiment_spec (which is only useful for FasterRCNN). Use this argument when DetectNet_v2 and FasterRCNN also set up the dataloader-based calibrator to leverage the training dataloader to calibrate the model. * :code:`--engine_file`: The path to the serialized TensorRT engine file. Note that this file is hardware specific and cannot be generalized across GPUs. Use this argument to quickly test your model accuracy using TensorRT on the host. As the TensorRT engine file is hardware specific, you cannot use this engine file for deployment unless the deployment GPU is identical to the training GPU. * :code:`--force_ptq`: A Boolean flag to force post-training quantization on the exported :code:`.etlt` model. .. Note:: When exporting a model that was trained with QAT enabled, the tensor scale factors to calibrate the activations are peeled out of the model and serialized to a TensorRT-readable cache file defined by the :code:`cal_cache_file` argument. However, the current version of QAT doesn’t natively support DLA int8 deployment on Jetson. To deploy this model on Jetson with DLA :code:`int8`, use the :code:`--force_ptq` flag to use TensorRT post-training quantization to generate the calibration cache file. Exporting a Model ^^^^^^^^^^^^^^^^^ Here's a sample command using the :code:`--cal_image_dir` option. .. code:: tlt ssd export -m $USER_EXPERIMENT_DIR/data/ssd/ssd_kitti_retrain_epoch12.tlt \ -o $USER_EXPERIMENT_DIR/data/ssd/ssd_kitti_retrain.int8.etlt \ -e $SPECS_DIR/ssd_kitti_retrain_spec.txt \ --key $KEY \ --cal_image_dir $USER_EXPERIMENT_DIR/data/KITTI/val/image_2 \ --data_type int8 \ --batch_size 8 \ --batches 10 \ --cal_data_file $USER_EXPERIMENT_DIR/data/ssd/cal.tensorfile \ --cal_cache_file $USER_EXPERIMENT_DIR/data/ssd/cal.bin \ --engine_file $USER_EXPERIMENT_DIR/data/ssd/detection.trt Deploying to Deepstream ----------------------- .. _deploying_to_deepstream_ssd: To run an SSD model in DeepStream, you need a label file and a DeepStream configuration file. In addition, you need to compile the TensorRT 7+ Open source software and DSSD bounding box parser for DeepStream. A DeepStream sample with documentation on how to run inference using the trained SSD models from TLT is provided here_ on GitHub. Prerequisites for SSD Model ^^^^^^^^^^^^^^^^^^^^^^^^^^^ SSD requires the batchTilePlugin, which is available in the TensorRT open source repo, but not in TensorRT 7.0. Detailed instructions to build TensorRT OSS can be found in `TensorRT Open Source Software (OSS)`_. .. _TensorRT Open Source Software (OSS): https://github.com/NVIDIA/TensorRT SSD requires custom bounding-box parsers that are not built in to the DeepStream SDK. The source code to build custom bounding box parsers for DSSD is available here_. The following instructions can be used to build the bounding box parser: .. _here: https://github.com/NVIDIA-AI-IOT/deepstream_tlt_apps/tree/master **Step 1**: Install git-lfs_ (git >= 1.8.2): .. _git-lfs: https://github.com/git-lfs/git-lfs/wiki/Installation .. Note:: git-lfs are needed to support downloading model files >5MB. .. code:: curl -s https://packagecloud.io/install/repositories/github/git-lfs/ script.deb.sh | sudo bash sudo apt-get install git-lfs git lfs install **Step 2**: Download the source code with HTTPS: .. code:: git clone -b release/tlt3.0 https://github.com/NVIDIA-AI-IOT/deepstream_tlt_apps **Step 3**: Build: .. code:: // or Path for DS installation export CUDA_VER=10.2 // CUDA version, e.g. 10.2 make This generates :code:`libnvds_infercustomparser_tlt.so` in the directory :code:`post_processor`. Label File ^^^^^^^^^^ The label file is a text file containing the names of the classes (with background) that the SSD model is trained to detect. The order in which the classes are listed in the label file must match the order in which the model predicts the output. During the training, TLT SSD will set the background as the first class, specify all other target class names in lower case, and sort them in alphabetical order. For example, if the :code:`dataset_config` is: .. code:: dataset_config { data_sources: { label_directory_path: "/path/to/train/labels" image_directory_path: "/path/to/train/images" } include_difficult_in_training: true target_class_mapping { key: "car" value: "car" } target_class_mapping { key: "person" value: "person" } target_class_mapping { key: "cyclist" value: "cyclist" } } Then the corresponding :code:`classification_lables.txt` file would look like this: .. code:: background bicycle car person DeepStream Configuration File ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ The detection model is typically used as a primary inference engine. It can also be used as a secondary inference engine. To run this model in the sample :code:`deepstream-app`, you must modify the existing :code:`config_infer_primary.txt` file to point to this model as well as the custom parser. .. image:: ../../content/dstream_deploy_options3.png **Option 1**: Integrate the model (:code:`.etlt`) directly in the DeepStream app. For this option, users will need to add the following parameters in the configuration file. The :code:`int8-calib-file` is only required for INT8 precision. .. code:: tlt-encoded-model= tlt-model-key= int8-calib-file= The tlt-encoded-model parameter points to the exported model (:code:`.etlt`) from TLT. The :code:`tlt-model-key` is the encryption key used during model export. **Option 2**: Integrate the TensorRT engine file with the DeepStream app. 1. Generate the TensorRT engine using :code:`tlt-converter`. 2. Once the engine file is generated successfully, modify the following parameters to use this engine with DeepStream: .. code:: model-engine-file= All other parameters are common between the two approaches. Add the label file generated above using following: .. code:: labelfile-path= For all the options, see the configuration file below. To learn more about the parameters, refer to the `DeepStream Development Guide`_. .. _DeepStream Development Guide: https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html .. code:: [property] gpu-id=0 net-scale-factor=1.0 offsets=103.939;116.779;123.68 model-color-format=1 labelfile-path= tlt-encoded-model= tlt-model-key= uff-input-dims=3;384;1248;0 uff-input-blob-name=Input batch-size=1 ## 0=FP32, 1=INT8, 2=FP16 mode network-mode=0 num-detected-classes=3 interval=0 gie-unique-id=1 is-classifier=0 #network-type=0 output-blob-names=NMS parse-bbox-func-name=NvDsInferParseCustomNMSTLT custom-lib-path= [class-attrs-all] threshold=0.3 roi-top-offset=0 roi-bottom-offset=0 detected-min-w=0 detected-min-h=0 detected-max-w=0 detected-max-h=0