I have been trying to convert a Yolo model into HEF so I can run it on the AI Hat+. After following the “hailo tutorial” notebooks I have made a quantized .har file. I am now trying to compile it and I am getting this error:
[info] Starting Hailo allocation and compilation flow
[error] Mapping Failed (allocation time: 15s)
No successful assignments: concat21 errors:
Agent infeasible
[error] Failed to produce compiled graph
[error] BackendAllocatorException: Compilation failed: No successful assignments: concat21 errors:
Agent infeasible
Can anyone help? I have been trying to get this to work for days.
Here are a few suggestions to help resolve the “Agent infeasible” error you encountered during compilation:
Check if concat21 is attempting to concatenate more than two layers at once. If so, try breaking it into multiple smaller concatenation steps. For example:
concat1 = concat([layer1, layer2])
concat2 = concat([concat1, layer3]) # Instead of trying to merge all three at once
Consider reducing the number of output channels in the layers preceding concat21. This can help if the concatenation is causing excessive memory usage.
Use the Hailo Profiler to identify resource bottlenecks:
hailo profile your_model.har
The generated report will help pinpoint the specific layer causing resource overuse.
If you continue to face issues, please provide more details about your model structure and the compiler logs. I’ll be happy to assist you further in debugging the problem.
