/dvs/git/dirty/gitlab-master_av/dw/sdk/tools/ingestion/apps/metadata/README-ingest_metadata.md

Go to the documentation of this file.

# Copyright (c) 2020, NVIDIA CORPORATION. All rights reserved.

@page dwx_metadata_ingestion_tool Metadata Ingestion Tool

### Description

This tool is available on the x86 Host System.

This is an executable which takes a binary sensor file as input, extracts the sensor metadata and outputs the result to a JSON file.

The tool currently supports Camera, CAN, GPS, IMU and Lidar binary file metadata extraction.

### Input
The input must be a DriveWorks-supported binary sensor file: *.raw, *.lraw, *.xraw or *.bin.

### Output
The output JSON file is filled with the sensor metadata. The output format of the metadata is described below for each sensor type.

### Usage
   cd install/tools/ingestion
   ./ingest_metadata --input=[path to input sensor file]
                     --output=[path to output JSON file]

### Output Format

##### CAN
- **codec** : *string*, "can"
- **fileVersion** : *int32_t*, file version of the recording

##### GPS
- **codec** : *string*, codec of the GPS device
- **fileVersion** : *int32_t*, file version of the recording

##### IMU
- **codec** : *string*, codec of the IMU device
- **fileVersion** : *int32_t*, file version of the recording

##### Video - Raw formats
- **codec** : *string*, { "image/dwraw", "image/dwlraw", "image/dwlraw_v2", "image/dwxraw" }
- **fileVersion** : *int32_t*, file version of the recording
- **codecData** : *string*, base64-encoding of a JSON string representing codec metadata
- **video** : video properties
   + **cameraType** : *string*, camera device type
   + **revision** : *uint8_t*, camera revision number
   + **frameRate** : *uint32_t*, frame rate
   + **dimensions** : video dimensions
      - **x** : *uint32_t*, width
      - **y** : *uint32_t*, height
   + **frameRegion** : frame region
      - **origin** : origin of the frame
         + **x** : *uint32_t*, x-coordinate of origin
         + **y** : *uint32_t*, y-coordinate of origin
      - **size** : size of the frame
         + **x** : *uint32_t*, width of the frame
         + **y** : *uint32_t*, height of the frame
   + **pixelFormat** : *string*, pixel format, e.g. "RCCB"
   + **pixelType** : *string*, e.g. "int16"
- **SIPLInfo** : SIPL header info (for v10 files and higher)
   + **cameraModuleName** : *string*
   + **cameraModuleDescription** : *string*
   + **sensorID** : *uint32_t*
   + **sensorName** : *string*
   + **sensorDescription** : *string*
- **xrawMetadata** : XRAW header info (for xraw videos only)
   + **packedFormat** : *uint32_t*
   + **packedWidth** : *uint32_t*
   + **packedHeight** : *uint32_t*
   + **packedGamma** : *float64_t*
   + **packedCodec** : *uint32_t*
   + **cameraType** : *uint32_t*

##### Video - MP4
- **codec** : *string*, "image/mp4"
- **video** : mp4 properties
   + **frameSize** : the frame dimensions
      - **x** : *uint32_t*, width
      - **y** : *uint32_t*, height
   + **mdatSectionSize** : *size_t*
   + **maxSampleSize** : *size_t*
   + **frameRate** : *float32_t*
   + **frameCount** : *size_t*
   + **syncEntriesCount** : *size_t*
   + **startTimestamp** : *int64_t*
   + **endTimestamp** : *int64_t*
   + **decodeFormat**
      - **codec** : *uint32_t*, 4 means H264
      - **chromaFormat** : *uint32_t*
      - **bitDepthLumaMinus8** : *uint8_t*
      - **codedWidth** : *uint32_t*
      - **codedHeight** : *uint32_t*
      - **displayWidth** : *uint32_t*
      - **displayHeight** : *uint32_t*

##### Lidar
- **codec** : *string*, name of the Lidar device
- **fileVersion** : *int32_t*, file version of the recording
- **codecData** : *string*, base64-string encoding of the bytes of device-specific metadata/calibration parameters; explained below
- **lidar** : lidar properties
   + **frameRate** : *uint32_t*, spins per second
   + **pointsPerFrame** : *uint32_t*, points per spin
   + **packetsPerFrame** : *uint32_t*, packets per spin

Bytes encoded in **codecData**:

1. for Ouster devices:
   - encoding version: 4 bytes, *uint32_t*
   - packets per spin: 4 bytes, *uint32_t*
   - ticks per azimuth: 4 bytes, *uint32_t*
   - beam angles: 2080 bytes, *Lidar::beamAngles*
2. for Velodyne devices:
   - encoding version: 4 bytes, *uint32_t*
   - packets per spin: 4 bytes, *uint32_t*
   - return mode: 4 bytes, *uint32_t*
   - calibration parameters: 64 X 44 bytes, *VelodyneCalibrationParams[64]*

The byte ordering for each of the data encoded above is little endian.