I was trying to use detectnet on a test-video, but the following error always occured:
5 objects detected
detected obj 0 class #3 (car) confidence=0.796387
bounding box 0 (1091.88, 521.72) (1271.88, 707.34) w=180.00 h=185.62
detected obj 1 class #6 (bus) confidence=0.539551
bounding box 1 (1091.88, 521.72) (1271.88, 707.34) w=180.00 h=185.62
detected obj 2 class #8 (truck) confidence=0.629395
bounding box 2 (448.75, 170.95) (543.12, 242.23) w=94.38 h=71.28
detected obj 3 class #8 (truck) confidence=0.556641
bounding box 3 (327.19, 179.12) (422.19, 246.97) w=95.00 h=67.85
detected obj 4 class #3 (car) confidence=0.727051
bounding box 4 (388.75, 219.38) (483.12, 271.05) w=94.38 h=51.68
[TRT] ------------------------------------------------
[TRT] Timing Report /usr/local/bin/networks/SSD-Mobilenet-v2/ssd_mobilenet_v2_coco.uff
[TRT] ------------------------------------------------
[TRT] Pre-Process CPU 0.12104ms CUDA 0.88266ms
[TRT] Network CPU 50.42432ms CUDA 48.20297ms
[TRT] Post-Process CPU 0.06875ms CUDA 0.06953ms
[cuda] cudaEventElapsedTime(&cuda_time, mEventsGPU[evt], mEventsGPU[evt+1])
[cuda] device not ready (error 600) (hex 0x258)
[cuda] /home/ben/jetson-inference/build/aarch64/include/jetson-inference/tensorNet.h:769
[TRT] Visualize CPU 0.53516ms CUDA 0.00000ms
[TRT] Total CPU 51.14927ms CUDA 49.15516ms
[TRT] ------------------------------------------------
nvbuf_utils: dmabuf_fd 1272 mapped entry NOT found
nvbuf_utils: NvReleaseFd Failed… Exiting…
The option to install SSD-MobileNet-v2 directly during build did not appear for me. Only the option to install pytorch was shown, which was skipped. I have used the latest version of jetpack available in the website.