How to use this with TF Serving? about cameratraps

[agentmorris]

Unfortunately, we (the MegaDetector team) don't have tons of TF Serving experience, so we won't be very helpful. Although it's a little out of date, the model is a pretty vanilla TF Object Detection API model, so you may have better luck on a forum/repo that is specific to TF serving, where you may be able to ask for examples of hosting TF 1.x object detection models via TF serving.

Sorry we don't have a more helpful answer here!

[supermoos]

Thanks anyway Dan! I did manage to figure out that the model just expects an input tensor named: “inputs” In case it helps anyone else. Best regards Lasse Moos -------------------- [phone]

…

Den 25. nov. 2021 kl. 17.10 skrev Dan Morris ***@***.***>:  Unfortunately, we (the MegaDetector team) don't have tons of TF Serving experience, so we won't be very helpful. Although it's a little out of date, the model is a pretty vanilla TF Object Detection API model, so you may have better luck on a forum/repo that is specific to TF serving, where you may be able to ask for examples of hosting TF 1.x object detection models via TF serving. Sorry we don't have a more helpful answer here! — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[agentmorris]

NP... and please, if you get it working, let us know! If there is a configuration file/example, we'd love to include that in the repo. Thanks.

[rbavery]

@supermoos Hi! I know this is old, but if you have an example, even scratch/old work I'd love to see this in a pr, email, or something else easy if you have the time to upload it! I'm currently hitting difficulties trying to reproduce locally what you accomplished with Google Vertex with the goal of standing up a Megadetector v4 model with AWS Sagemaker. I've tried v1.15, v1.14 and v1.13 but no luck, I get graph errors like this (for tf v1.15):

tnc-ca-geo/animl-ml#70

[rbavery]

I've addressed some of the issues with just getting the model server up. It seems like we need this tree structure

models/export-tfserve-empty-vardir
├── code
│   ├── inference.py
│   └── requirements.txt
└── Servo
    └── 1
        ├── saved_model.pb
        └── variables/

and nothing in the variables directory (including the checkpoint linked in the readme for v4.1 resulted in an error

and now I'm at where this issue seemed to get where I have a payload that looks like

'{"inputs": [{"b64": "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"}]}'

when I try to post this to "http://localhost:8501/v1/models/model/predict"

I get

r.text
'{ "error": "Malformed request: POST /v1/models/model/predict" }'

r.status_code
400

I also tried this with instances as the field as I've seen in the tfserving docs but same result. working on getting a more informative error message from the model server...