Sacha

Example in C

2

Is there any example to set up the context required to use this backend for decoding in C? I compared this to CUVID and it appears like it is missing...

Install the config

3

Install config for cpuinfo so that it can found with find_package(cpuinfo) and used with target_link_libraries(... cpuinfo::cpuinfo)

Increase threshold for NMS

3

The threshold given to NMS (0.3) is far too low and includes many false positives. If someone wanted to see them, they can set the value themselves but I think...

Improves faceboxes test scores on major benchmarks PASCAL: 97.05% -> 97.28% AFW: 98.55% -> 98.83% I really didn't like these magic numbers for each dataset, so instead I targeted a...

Accuracy with classes

5

Hi, I often see 'Giraffe' or other wild animals appear around ~70% when inferencing on images. I would just like to confirm I am using the correct class set (COCO_CATEGORIES...

Thanks for your work decompiling this app and showing the JavaScript and SDK strings. I took interest in your analysis but it left me with more questions than answers. Some...

Any prebuilt weights available?

5

I notice you only went to epoch 22, which is still stage 1. Are you still training this?

Not normalised -1 to 1?

2

Ideally the input data is normalised between -1 and 1 because fp32 and fp16 have better precision in that range. Is there a reason why it is commented out in...

Batch size in config not used

1

The batch size in config.py is not used for anything (instead it used the -b option). This confused me initially. Maybe it could change the default value.

Why is the centre... off-centre? This surely creates bias in the training.