- Open Hardware & Model
- Open Hardware & Closed Model
- Closed Hardware & Open Model
- Closed Hardware & Model
Thereby, "open" and "closed" refer to the option of using different hardware platforms or ML models. Note that the submission category "Closed Hardware & Model" aims to compare different pre- and postprocessing techniques and hardware components.
Use Case | Task | Dataset | Reference ML Model | Reference HW |
---|---|---|---|---|
A | Anomaly Detection | SMAP & MSL | N-BEATS | Xilinx Zynq Ultrascale+ ZCU102 |
B | Radio Classification | RadioML 2018.01A | ResNet 1d | Xilinx Zynq Ultrascale+ ZCU102 |
C | Multispectral Object Detection | OroraTech Wildfire Dataset | VGG 16 | Xilinx Zynq Ultrascale+ ZCU102 |
D | Image Classification Heavy | EuroSAT | DenseNet 161 | Xilinx Zynq Ultrascale+ ZCU102 |
E | Image Classification Light | Aibus Ship Detection | Mobilenet | Xilinx Zynq Ultrascale+ ZCU102 |
F | Image Object Detection | Airbus Aircraft Detection | Yolov4 | Xilinx Zynq Ultrascale+ ZCU102 |
G | Image Segmentation | Aibus Ship Detection | ResNet 50 | Xilinx Zynq Ultrascale+ ZCU102 |
You are welcome to participate in the benchmark by submitting your results on any of the applications. For a new submission, two files need to be prepared:
configuration.json
{
"task": {
"use_case": "",
"dataset": "",
"target_metrics": [],
"inference_scenario": "",
"submission_category": ""
},
"system": {
"device": "",
"hardware_image": "",
"os": "",
"framework": "",
"application_type": "",
"relevant_drivers": [],
"power_monitors": [],
"reproducibility_guidelines": ""
},
"model": {
"name": "",
"architecture": ""
},
"link_to_project": ""
}
metrics.json, e.g.
{
"accuracy": "",
"throughput": "",
"energy": "",
"peak_power": "",
"idle_power": ""
}
Thereby, the task use case, dataset and target metrics are fixed for the existing applications and examples can be found for the reference implementations. The inference scenario refers to either "batch" or "stream" and the submission category has to be chosen as one or more of the 4 categories described above, specifying whether the hardware or model are allowed to be varied. Make sure to add a link to your project!
Once these two files have been prepared, a submission can be made via a pull request:
- Clone directory and create new branch with <unique name>
- Create folder in /submissions/<task>/<submission category>/<unique name>
- Add configuration.json and metrics.json
- Copy the newly created folder to /website/_data/submissions/<task>/<submission category>/<unique name>
- File a pull request to merge your branch into the main branch
- Koch, A., Dax, G., Petry, M., Gomez, H., Raoofy, A., Saroliya, U., Ghiglione, M., Furano, G., Werner, M., Trinitis, C., & Langer, M. (2023). Reference Implementations for Machine Learning Application Benchmark. 2023 European Data Handling & Data Processing Conference (EDHPC), Juan Les Pins, France, 2023, pp. 1-3, https://doi.org/10.23919/EDHPC59100.2023.10396582
- Koch, A., Petry, M., Ghiglione, M., Raoofy, A., Dax, G., Furano, G., Werner, M., Trinitis, C., & Langer, M. (2023). Machine Learning Application Benchmark. 20th ACM International Conference on Computing Frontiers (CF ’23), May 9–11, 2023, Bologna, Italy. https://doi.org/10.1145/3587135.3592769
- Dax, G., Nagarajan, S., Li, H., & Werner, M. (2022). Compression Supports Spatial Deep Learning. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing (JSTARS). https://doi.org/10.1109/JSTARS.2022.3226563
- Dax, G., & Werner, M. (2022). The Role of Compression in Spatial Computing. PhD Colloquium of the Deutsche Geodätische Kommission, Section on Geoinformatics. https://www.bgd.ed.tum.de/pdf/2022_phdColoquiumDGK_Dax.pdf
- Denizoglu, D. G., Dax, G., Nagarajan, S., Zhang, N., & Werner, M. (2022). Global Active Fire Detection – Towards a SAR-enabled Multi-Sensor Global Monitoring System. Living Planet Symposium 2022. https://www.bgd.ed.tum.de/pdf/2022_wildfireS2_Gaye.pdf
- Raoofy, A., Dax, G., Serra, V., Ghiglione, M., Werner, M., & Trinitis, C. (2022). Benchmarking and Feasibility Aspects of Machine Learning in Space Systems. Proceedings of the 19th ACM International Conference on Computing Frontiers (CF’22). https://doi.org/10.1145/3528416.3530986
- Ghiglione, M., Serra, V., Raoofy, A., Dax, G., Trinitis, C., Werner, M., Schulz, M., & Furano, G. (2022). Survey of frameworks for inference of neural networks in space data system. Data Systems in Aerospace (DASIA). Eurospace. https://www.bgd.ed.tum.de/pdf/2022_cnnInSpace_Ghiglione.pdf
- Ghiglione, M., Raoofy, A., Dax, G., Furano, G., Wiest, R., Trinitis, C., Werner, M., Schulz, M., & Langer, M. (2021). Machine Learning Application Benchmark for In-Orbit On-Board Data Processing. European Workshop on On-Board Data Processing. https://zenodo.org/record/5520877/files/05.04_OBDP2021_Ghiglione.pdf
- Raoofy, A., Dax, G., Ghiglione, M., Langer, M., Trinitis, C., Werner, M., & Schulz, M. (2021). Benchmarking Machine Learning Inference in FPGA-based Accelerated Space Applications. Proceedings of the Workshop on Benchmarking Machine Learning Workloads Co-Located with IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS). https://www.bgd.ed.tum.de/pdf/2021_fpgabenchmarkpositionpaper_Raoofy.pdf