Access S3 Stored Images#

The DL Streamer Pipeline Server stores processed images in the SeaweedFS S3 bucket. This guide explains how to access and verify these stored images.

Overview#

All images processed by the DL Streamer Pipeline Server are stored in the S3 bucket named dlstreamer-pipeline-results. The images are stored in the weld-defect-classification/ directory and named using their unique img_handle identifier.

Viewing Vision Metadata in InfluxDB#

The DL Streamer Pipeline Server generates vision metadata for each processed frame. This metadata is stored in InfluxDB.

Accessing Vision Metadata#

  1. Connect to InfluxDB container:

    docker exec -it ia-influxdb bash

    NOTE: Use kubectl exec -it <influxdb-pod-name> -n <namespace> -- /bin/bash for the helm deployment where for replace with namespace name where the application was deployed and for replace with InfluxDB pod name.

  2. Query the vision metadata:

    # For below command, the INFLUXDB_USERNAME and INFLUXDB_PASSWORD needs to be fetched from `.env` file
influx -username <username> -password <password>
USE datain
SHOW MEASUREMENTS

# View vision detection results
SELECT * FROM "vision-weld-classification-results"

NOTE: You may see the error There was an error writing history file: open /.influx_history: read-only file system in the InfluxDB shell. This is harmless and does not affect functionality.

Accessing Stored Images using SeaweedFS Filer Web Interface#

Access the SeaweedFS Filer interface in your web browser:

https://<host_ip>:3000/image-store/buckets/dlstreamer-pipeline-results/weld-defect-classification/

Note: Use link https://<host_ip>:30001/image-store/buckets/dlstreamer-pipeline-results/weld-defect-classification/ to access the SeaweedFS Filer interface for the Helm deployment.

Images are organized by their img_handle identifier. Browse the directory to locate specific images, then click to view the image.

Mapping Vision Metadata to Stored Images#

Follow these steps to correlate detection events in InfluxDB with stored images:

  1. Query InfluxDB to retrieve vision metadata:

    SELECT * FROM "vision-weld-classification-results"

  2. Note the img_handle from the query results (e.g., X7TINNVPNX).

  3. Navigate to the Filer interface:

    http://<host_ip>:8887/buckets/dlstreamer-pipeline-results/weld-defect-classification/

  4. Locate and open the file matching the img_handle (e.g., X7TINNVPNX.jpg).

Note: All data stored in SeaweedFS and InfluxDB is non-persistent and will be lost on container/pod restart.