gaopeng-eugene

Do you plan to include COCO data training into your project

Any idea on how to improve the performance?

How to restore learning rate for Adam optimiser?

It seems that the problem is in lib/dataset/coco.py line 371: results = pool.map(coco_results_one_category_kernel, data_pack)

My solution right now is folllowing : coco.py line 371 : replace pool = mp.Pool(mp.cpu_count()) results = pool.map(coco_results_one_category_kernel, data_pack) pool.close() pool.join() with results = [] for i in range(len(data_pack)): results.append(coco_results_one_category_kernel(data_pack[i]))...

no

Of course, my code running in single process while the original code running in multiple process. My guess is there is deadlock in the original code.

Is it possible for you to share the performance on COCO test and val?

Where is the describe function in your test code?

can you provide the trained_segment_weights.pkl file?